Just a quick bit about Snorkel. We're backed by top tier investors (https://snorkel.ai/company/) and just raised our Series C last year, putting us in a really well strong position to continue building Snorkel Flow and our business. Our customers include some of the world's largest enterprises \u2014 top US banks, Memorial Sloan Kettering Cancer Center, Chubb, F500 biotechs, and more (https://snorkel.ai/case-studies/) \u2014 and we're empowering their data science teams to build high quality models in just hours or days instead of weeks or months. But we're really just getting started on our journey, and there are so many diverse opportunities for having impact across the company.
As a senior front-end engineer, you'll build amazing interfaces that will empower developers and non-developers alike to rapidly build, manage, and analyze ML pipelines in Snorkel Flow.
Front-end is incredibly important to our mission here at Snorkel. The intuitive\u2014and as some users have called them, fun\u2014workflows we've created to help data scientists and subject matter experts quickly build high quality training data sets and ML models are what set us apart. Right now, we're looking for leadership across some major initiatives focusing on \n* performance and snappiness (streaming server-side rendering, SWR, React Suspense etc.);\n* deepening our design system and component library (Storybook, Chromatic); and \n* general front-end codebase architecture to make sure we can keep scaling and moving fast.
Absolutely no need to have experience or deep familiarity with ML/AI, but there's no better place and no better team to work with if you're interested in learning what it takes to build production ML systems.
## Main Responsibilities\n* Design and develop key interfaces of Snorkel Flow, including a novel IDE for programmatic labeling, data analysis, and visualization tools\n* Advocate and help define standards for the best practices in performance, product quality, security and user experience across the entire application\n* Provide mentorship and foster a growth and collaborative mindset for all members of your team.\n* Collaborate with cross functional teams on developing large scale products\n* Proactively identify technical debt and implement solutions to improve product iteration
## Preferred Qualifications\n* 5+ years of professional experience with modern React (ES7, React hooks, TypeScript)\n* Strong technical communication skills and initiative\n* Experience leading front-end projects / features\n* Product engineering experience building full-stack web applications\n* Ability to prototype and iterate quickly around sophisticated user workflows\n* Obsession with front-end development best practices, including designing unit and end-to-end tests\n* Ability to work in a fast-paced, startup environment
You can find more information on our job posting (https://boards.greenhouse.io/snorkelai/jobs/4136833004), in this FAQ (https://snorkel.ai/faq/), and on our website (https://snorkel.ai/join-us/).
If interested, email rakesh.guha@snorkel.ai and sharyl.sun@snorkel.ai."},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Snorkel AI is looking for core front end engineers"}},"_tags":["story","author_guharakesh","story_31899407","ask_hn"],"author":"guharakesh","created_at":"2022-06-27T19:17:24Z","created_at_i":1656357444,"num_comments":0,"objectID":"31899407","points":2,"story_id":31899407,"story_text":"Snorkel AI | https://snorkel.ai/ | Senior Frontend Engineer | Full-time | ONSITE | SF Bay Area
Just a quick bit about Snorkel. We're backed by top tier investors (https://snorkel.ai/company/) and just raised our Series C last year, putting us in a really well strong position to continue building Snorkel Flow and our business. Our customers include some of the world's largest enterprises \u2014 top US banks, Memorial Sloan Kettering Cancer Center, Chubb, F500 biotechs, and more (https://snorkel.ai/case-studies/) \u2014 and we're empowering their data science teams to build high quality models in just hours or days instead of weeks or months. But we're really just getting started on our journey, and there are so many diverse opportunities for having impact across the company.
As a senior front-end engineer, you'll build amazing interfaces that will empower developers and non-developers alike to rapidly build, manage, and analyze ML pipelines in Snorkel Flow.
Front-end is incredibly important to our mission here at Snorkel. The intuitive\u2014and as some users have called them, fun\u2014workflows we've created to help data scientists and subject matter experts quickly build high quality training data sets and ML models are what set us apart. Right now, we're looking for leadership across some major initiatives focusing on \n* performance and snappiness (streaming server-side rendering, SWR, React Suspense etc.);\n* deepening our design system and component library (Storybook, Chromatic); and \n* general front-end codebase architecture to make sure we can keep scaling and moving fast.
Absolutely no need to have experience or deep familiarity with ML/AI, but there's no better place and no better team to work with if you're interested in learning what it takes to build production ML systems.
## Main Responsibilities\n* Design and develop key interfaces of Snorkel Flow, including a novel IDE for programmatic labeling, data analysis, and visualization tools\n* Advocate and help define standards for the best practices in performance, product quality, security and user experience across the entire application\n* Provide mentorship and foster a growth and collaborative mindset for all members of your team.\n* Collaborate with cross functional teams on developing large scale products\n* Proactively identify technical debt and implement solutions to improve product iteration
## Preferred Qualifications\n* 5+ years of professional experience with modern React (ES7, React hooks, TypeScript)\n* Strong technical communication skills and initiative\n* Experience leading front-end projects / features\n* Product engineering experience building full-stack web applications\n* Ability to prototype and iterate quickly around sophisticated user workflows\n* Obsession with front-end development best practices, including designing unit and end-to-end tests\n* Ability to work in a fast-paced, startup environment
You can find more information on our job posting (https://boards.greenhouse.io/snorkelai/jobs/4136833004), in this FAQ (https://snorkel.ai/faq/), and on our website (https://snorkel.ai/join-us/).
If interested, email rakesh.guha@snorkel.ai and sharyl.sun@snorkel.ai.","title":"Snorkel AI is looking for core front end engineers","updated_at":"2024-09-20T11:31:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jerodsanto"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Getting in the Flow with Snorkel AI [audio]"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://changelog.com/practicalai/117"}},"_tags":["story","author_jerodsanto","story_25498143"],"author":"jerodsanto","created_at":"2020-12-21T19:00:03Z","created_at_i":1608577203,"num_comments":0,"objectID":"25498143","points":2,"story_id":25498143,"title":"Getting in the Flow with Snorkel AI [audio]","updated_at":"2024-09-20T07:34:35Z","url":"https://changelog.com/practicalai/117"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"blopeur"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Snorkel AI: Putting Data First in ML Development"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://www.snorkel.ai/07-14-2020-snorkel-ai-launch"}},"_tags":["story","author_blopeur","story_23838310"],"author":"blopeur","created_at":"2020-07-14T21:40:42Z","created_at_i":1594762842,"num_comments":0,"objectID":"23838310","points":2,"story_id":23838310,"title":"Snorkel AI: Putting Data First in ML Development","updated_at":"2024-09-20T06:35:22Z","url":"https://www.snorkel.ai/07-14-2020-snorkel-ai-launch"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"wavelander"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Enterprise Alignment from Snorkel AI"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://snorkel.ai/introducing-enterprise-alignment-path-to-superalignment/"}},"_tags":["story","author_wavelander","story_40416941"],"author":"wavelander","created_at":"2024-05-20T16:02:41Z","created_at_i":1716220961,"num_comments":0,"objectID":"40416941","points":1,"story_id":40416941,"title":"Enterprise Alignment from Snorkel AI","updated_at":"2024-09-20T17:00:57Z","url":"https://snorkel.ai/introducing-enterprise-alignment-path-to-superalignment/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jerodsanto"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Getting in the Flow with Snorkel AI (Transcript)"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://changelog.com/practicalai/117#transcript"}},"_tags":["story","author_jerodsanto","story_25520094"],"author":"jerodsanto","created_at":"2020-12-23T17:49:24Z","created_at_i":1608745764,"num_comments":0,"objectID":"25520094","points":1,"story_id":25520094,"title":"Getting in the Flow with Snorkel AI (Transcript)","updated_at":"2024-09-20T07:34:56Z","url":"https://changelog.com/practicalai/117#transcript"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"someguy101010"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Hey HN, we're excited to share Cua-Bench ( https://github.com/trycua/cua ), an open-source framework for evaluating and training computer-use agents across different environments.
Computer-use agents show massive performance variance across different UIs\u2014an agent with 90% success on Windows 11 might drop to 9% on Windows XP for the same task. The problem is OS themes, browser versions, and UI variations that existing benchmarks don't capture.
The existing benchmarks (OSWorld, Windows Agent Arena, AndroidWorld) were great but operated in silos\u2014different harnesses, different formats, no standardized way to test the same agent across platforms. More importantly, they were evaluation-only. We needed environments that could generate training data and run RL loops, not just measure performance.\nCua-Bench takes a different approach: it's a unified framework that standardizes environments across platforms and supports the full agent development lifecycle\u2014benchmark, train, deploy.
With Cua-Bench, you can:
- Evaluate agents across multiple benchmarks with one CLI (native tasks + OSWorld + Windows Agent Arena adapters)
- Test the same agent on different OS variations (Windows 11/XP/Vista, macOS themes, Linux, Android via QEMU)
- Generate new tasks from natural language prompts
- Create simulated environments for RL training (shell apps like Spotify, Slack with programmatic rewards)
- Run oracle validations to verify environments before agent evaluation
- Monitor agent runs in real-time with traces and screenshots
All of this works on macOS, Linux, Windows, and Android, and is self-hostable.
To get started:
Install cua-bench:
% pip install cua-bench
Run a basic evaluation:
% cb run dataset datasets/cua-bench-basic --agent demo
Open the monitoring dashboard:
% cb run watch <run_id>
For parallelized evaluations across multiple workers:
% cb run dataset datasets/cua-bench-basic --agent your-agent --max-parallel 8
Want to test across different OS variations? Just specify the environment:
% cb run task slack_message --agent your-agent --env windows_xp
% cb run task slack_message --agent your-agent --env macos_sonoma
Generate new tasks from prompts:
% cb task generate "book a flight on kayak.com"
Validate environments with oracle implementations:
% cb run dataset datasets/cua-bench-basic --oracle
The simulated environments are particularly useful for RL training\u2014they're HTML/JS apps that render across 10+ OS themes with programmatic reward verification. No need to spin up actual VMs for training loops.
We're seeing teams use Cua-Bench for:
- Training computer-use models on mobile and desktop environments
- Generating large-scale training datasets (working with labs on millions of screenshots across OS variations)
- RL fine-tuning with shell app simulators
- Systematic evaluation across OS themes and browser versions
- Building task registries (collaborating with Snorkel AI on task design and data curation, similar to their Terminal-Bench work)
Cua-Bench is 100% open-source under the MIT license. We're actively developing it as part of Cua (https://github.com/trycua/cua), our Computer Use Agent SDK, and we'd love your feedback, bug reports, or feature ideas.
GitHub: https://github.com/trycua/cua
Docs: https://cua.ai/docs/cuabench
Technical Report: https://cuabench.ai
We'll be here to answer any technical questions and look forward to your comments!"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: Cua-Bench \u2013 a benchmark for AI agents in GUI environments"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/trycua/cua"}},"_tags":["story","author_someguy101010","story_46768906","show_hn"],"author":"someguy101010","children":[46797173,46800251,46800556,46804011,46804855],"created_at":"2026-01-26T17:46:22Z","created_at_i":1769449582,"num_comments":8,"objectID":"46768906","points":40,"story_id":46768906,"story_text":"Hey HN, we're excited to share Cua-Bench ( https://github.com/trycua/cua ), an open-source framework for evaluating and training computer-use agents across different environments.
Computer-use agents show massive performance variance across different UIs\u2014an agent with 90% success on Windows 11 might drop to 9% on Windows XP for the same task. The problem is OS themes, browser versions, and UI variations that existing benchmarks don't capture.
The existing benchmarks (OSWorld, Windows Agent Arena, AndroidWorld) were great but operated in silos\u2014different harnesses, different formats, no standardized way to test the same agent across platforms. More importantly, they were evaluation-only. We needed environments that could generate training data and run RL loops, not just measure performance.\nCua-Bench takes a different approach: it's a unified framework that standardizes environments across platforms and supports the full agent development lifecycle\u2014benchmark, train, deploy.
With Cua-Bench, you can:
- Evaluate agents across multiple benchmarks with one CLI (native tasks + OSWorld + Windows Agent Arena adapters)
- Test the same agent on different OS variations (Windows 11/XP/Vista, macOS themes, Linux, Android via QEMU)
- Generate new tasks from natural language prompts
- Create simulated environments for RL training (shell apps like Spotify, Slack with programmatic rewards)
- Run oracle validations to verify environments before agent evaluation
- Monitor agent runs in real-time with traces and screenshots
All of this works on macOS, Linux, Windows, and Android, and is self-hostable.
To get started:
Install cua-bench:
% pip install cua-bench
Run a basic evaluation:
% cb run dataset datasets/cua-bench-basic --agent demo
Open the monitoring dashboard:
% cb run watch <run_id>
For parallelized evaluations across multiple workers:
% cb run dataset datasets/cua-bench-basic --agent your-agent --max-parallel 8
Want to test across different OS variations? Just specify the environment:
% cb run task slack_message --agent your-agent --env windows_xp
% cb run task slack_message --agent your-agent --env macos_sonoma
Generate new tasks from prompts:
% cb task generate "book a flight on kayak.com"
Validate environments with oracle implementations:
% cb run dataset datasets/cua-bench-basic --oracle
The simulated environments are particularly useful for RL training\u2014they're HTML/JS apps that render across 10+ OS themes with programmatic reward verification. No need to spin up actual VMs for training loops.
We're seeing teams use Cua-Bench for:
- Training computer-use models on mobile and desktop environments
- Generating large-scale training datasets (working with labs on millions of screenshots across OS variations)
- RL fine-tuning with shell app simulators
- Systematic evaluation across OS themes and browser versions
- Building task registries (collaborating with Snorkel AI on task design and data curation, similar to their Terminal-Bench work)
Cua-Bench is 100% open-source under the MIT license. We're actively developing it as part of Cua (https://github.com/trycua/cua), our Computer Use Agent SDK, and we'd love your feedback, bug reports, or feature ideas.
GitHub: https://github.com/trycua/cua
Docs: https://cua.ai/docs/cuabench
Technical Report: https://cuabench.ai
We'll be here to answer any technical questions and look forward to your comments!","title":"Show HN: Cua-Bench \u2013 a benchmark for AI agents in GUI environments","updated_at":"2026-03-05T23:26:06Z","url":"https://github.com/trycua/cua"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"4thabang"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Our AI recruitment pipeline was auto-rejecting anyone who'd worked at companies founded after 2023. It didn't recognize names like Harvey or Snorkel AI, or didn't realize how important they'd become because of training data cutoffs.
We had traces, evals, Langfuse dashboards - everything looked fine - but we kept finding failures we should have caught earlier.
The pattern kept repeating:
- ship an improvement\n- it works for a while\n- hit an edge case that breaks it\n- don't notice until we've lost good candidates
That's when we realized - the problem wasn't just our recruitment pipeline - almost every AI product has blind spots that evals miss.
So we built Verse, a tool that surfaces issues directly from real AI interactions - whether that's candidates talking to your recruitment pipeline, users interacting with your agent, or any AI making decisions.
Instead of relying solely on evals, we cluster conversations, identify the key ones to review, and flag the ones that show failure patterns. We use OpenTelemetry for trace ingestion, so it's compatible with Langfuse, Langsmith, Braintrust, and other AI observability tools - you can add it right alongside your existing setup.
I'm posting this because I'm curious whether other teams are hitting the same wall. If you want, I'm happy to audit your AI implementation for free and show you where things commonly break - even if you never use Verse.
Happy to answer any technical questions."},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: Verse AI \u2013 Catch the AI failures your evals miss"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://tryverse.ai/"}},"_tags":["story","author_4thabang","story_45914386","show_hn"],"author":"4thabang","created_at":"2025-11-13T13:00:01Z","created_at_i":1763038801,"num_comments":0,"objectID":"45914386","points":5,"story_id":45914386,"story_text":"Our AI recruitment pipeline was auto-rejecting anyone who'd worked at companies founded after 2023. It didn't recognize names like Harvey or Snorkel AI, or didn't realize how important they'd become because of training data cutoffs.
We had traces, evals, Langfuse dashboards - everything looked fine - but we kept finding failures we should have caught earlier.
The pattern kept repeating:
- ship an improvement\n- it works for a while\n- hit an edge case that breaks it\n- don't notice until we've lost good candidates
That's when we realized - the problem wasn't just our recruitment pipeline - almost every AI product has blind spots that evals miss.
So we built Verse, a tool that surfaces issues directly from real AI interactions - whether that's candidates talking to your recruitment pipeline, users interacting with your agent, or any AI making decisions.
Instead of relying solely on evals, we cluster conversations, identify the key ones to review, and flag the ones that show failure patterns. We use OpenTelemetry for trace ingestion, so it's compatible with Langfuse, Langsmith, Braintrust, and other AI observability tools - you can add it right alongside your existing setup.
I'm posting this because I'm curious whether other teams are hitting the same wall. If you want, I'm happy to audit your AI implementation for free and show you where things commonly break - even if you never use Verse.
Happy to answer any technical questions.","title":"Show HN: Verse AI \u2013 Catch the AI failures your evals miss","updated_at":"2026-03-05T22:58:23Z","url":"https://tryverse.ai/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"matt_d"},"title":{"matchLevel":"none","matchedWords":[],"value":"Senior SWE-Bench: open-source benchmark that assesses agents as senior engineers"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://senior-swe-bench.snorkel.ai/"}},"_tags":["story","author_matt_d","story_48755928"],"author":"matt_d","children":[48756144,48756179,48756211,48756301,48756349,48756470,48756667,48756760,48756761,48756800,48756823,48756842,48757061,48757602,48757795,48758157,48759519,48759786,48760877,48762698,48763608,48764183,48765059,48765498,48765545,48767668,48770074],"created_at":"2026-07-02T02:55:16Z","created_at_i":1782960916,"num_comments":120,"objectID":"48755928","points":186,"story_id":48755928,"title":"Senior SWE-Bench: open-source benchmark that assesses agents as senior engineers","updated_at":"2026-07-25T05:14:09Z","url":"https://senior-swe-bench.snorkel.ai/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"vincentschen"},"story_text":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Today, we're launching the Open Benchmarks Grants: a $3M commitment to fund open-source and academic teams building benchmarks for AI agents. In partnership with HuggingFace, PrimeIntellect, FactoryHQ, Together, Harbor, and PyTorch, the grants provide funding, data development support, and research collaboration.
Our ability to measure AI has been outpaced by our ability to develop it, and we believe this evaluation gap is one of the most important problems in AI. Open benchmarks are one of the most important levers for advancing AI safely and responsibly\u2014but the academic and open-source teams driving them often hit resource constraints, especially in the face of the exponentially expanding complexity of what tomorrow\u2019s benchmarks need to cover.
We think the next wave of benchmarks needs to push on three axes:\n- Environment complexity - How realistic is the operating environment?\n- Autonomy horizon - How far can an agent operate independently? We need to measure \n- Output complexity - How sophisticated is the work product?
Happy to answer questions about the grants, the framework, and would love to hear more about what you\u2019re building!"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: Open Benchmarks Grants\u2013 a $3M commitment to close the AI eval gap"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://benchmarks.snorkel.ai/closing-the-evaluation-gap-in-agentic-ai/"}},"_tags":["story","author_vincentschen","story_46979278","show_hn"],"author":"vincentschen","created_at":"2026-02-11T19:06:18Z","created_at_i":1770836778,"num_comments":0,"objectID":"46979278","points":6,"story_id":46979278,"story_text":"Today, we're launching the Open Benchmarks Grants: a $3M commitment to fund open-source and academic teams building benchmarks for AI agents. In partnership with HuggingFace, PrimeIntellect, FactoryHQ, Together, Harbor, and PyTorch, the grants provide funding, data development support, and research collaboration.
Our ability to measure AI has been outpaced by our ability to develop it, and we believe this evaluation gap is one of the most important problems in AI. Open benchmarks are one of the most important levers for advancing AI safely and responsibly\u2014but the academic and open-source teams driving them often hit resource constraints, especially in the face of the exponentially expanding complexity of what tomorrow\u2019s benchmarks need to cover.
We think the next wave of benchmarks needs to push on three axes:\n- Environment complexity - How realistic is the operating environment?\n- Autonomy horizon - How far can an agent operate independently? We need to measure \n- Output complexity - How sophisticated is the work product?
Happy to answer questions about the grants, the framework, and would love to hear more about what you\u2019re building!","title":"Show HN: Open Benchmarks Grants\u2013 a $3M commitment to close the AI eval gap","updated_at":"2026-03-05T23:34:57Z","url":"https://benchmarks.snorkel.ai/closing-the-evaluation-gap-in-agentic-ai/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"robiriondo"},"title":{"matchLevel":"none","matchedWords":[],"value":"Using few-shot learning language models as weak supervision"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://snorkel.ai/few-shot-learning-large-language-models/"}},"_tags":["story","author_robiriondo","story_31262775"],"author":"robiriondo","created_at":"2022-05-04T16:36:17Z","created_at_i":1651682177,"num_comments":0,"objectID":"31262775","points":3,"story_id":31262775,"title":"Using few-shot learning language models as weak supervision","updated_at":"2024-09-20T11:05:42Z","url":"https://snorkel.ai/few-shot-learning-large-language-models/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"debug_overflow"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"7B model from Snorkel tops Alpaca Eval 2.0 leaderboard"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://huggingface.co/snorkelai/Snorkel-Mistral-PairRM-DPO"}},"_tags":["story","author_debug_overflow","story_39122883"],"author":"debug_overflow","children":[39122983,39123588],"created_at":"2024-01-24T21:32:19Z","created_at_i":1706131939,"num_comments":2,"objectID":"39122883","points":2,"story_id":39122883,"title":"7B model from Snorkel tops Alpaca Eval 2.0 leaderboard","updated_at":"2024-09-20T16:11:51Z","url":"https://huggingface.co/snorkelai/Snorkel-Mistral-PairRM-DPO"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"vinhnx"},"title":{"matchLevel":"none","matchedWords":[],"value":"Evaluating Coding Agents with Terminal-Bench 2.0"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://snorkel.ai/blog/evaluating-coding-agent-capabilities-with-terminal-bench-snorkels-role-in-building-the-next-generation-benchmark/"}},"_tags":["story","author_vinhnx","story_45469851"],"author":"vinhnx","created_at":"2025-10-04T01:55:35Z","created_at_i":1759542935,"num_comments":0,"objectID":"45469851","points":2,"story_id":45469851,"title":"Evaluating Coding Agents with Terminal-Bench 2.0","updated_at":"2026-03-05T22:49:55Z","url":"https://snorkel.ai/blog/evaluating-coding-agent-capabilities-with-terminal-bench-snorkels-role-in-building-the-next-generation-benchmark/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"willhang"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"Redesigning Snorkel\u2019s interactive machine learning systems"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://snorkel.ai/redesigning-snorkel-interactive-machine-learning-systems/"}},"_tags":["story","author_willhang","story_35993159"],"author":"willhang","created_at":"2023-05-18T19:49:13Z","created_at_i":1684439353,"num_comments":0,"objectID":"35993159","points":2,"story_id":35993159,"title":"Redesigning Snorkel\u2019s interactive machine learning systems","updated_at":"2024-09-20T14:07:26Z","url":"https://snorkel.ai/redesigning-snorkel-interactive-machine-learning-systems/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"robiriondo"},"title":{"matchLevel":"none","matchedWords":[],"value":"Liger: Fusing foundation model embeddings and weak supervision"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://snorkel.ai/liger-fusing-model-embeddings-weak-supervision/"}},"_tags":["story","author_robiriondo","story_31330244"],"author":"robiriondo","created_at":"2022-05-10T18:01:05Z","created_at_i":1652205665,"num_comments":0,"objectID":"31330244","points":2,"story_id":31330244,"title":"Liger: Fusing foundation model embeddings and weak supervision","updated_at":"2024-09-20T11:05:34Z","url":"https://snorkel.ai/liger-fusing-model-embeddings-weak-supervision/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"vincentschen"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Frontier AI Is Accelerating. Open Benchmarks Need to Keep Up"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://benchmarks.snorkel.ai/frontier-ai-is-accelerating-open-benchmarks-need-to-keep-up/"}},"_tags":["story","author_vincentschen","story_49998562"],"author":"vincentschen","created_at":"2026-10-07T20:45:06Z","created_at_i":1791405906,"num_comments":0,"objectID":"49998562","points":1,"story_id":49998562,"title":"Frontier AI Is Accelerating. Open Benchmarks Need to Keep Up","updated_at":"2026-10-07T20:47:51Z","url":"https://benchmarks.snorkel.ai/frontier-ai-is-accelerating-open-benchmarks-need-to-keep-up/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rjurney"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Hand labeling is the past. The future is NoLabel AI with Snorkel"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://www.kdnuggets.com/2020/02/hand-labeling-past-future-nolabel-ai.html"}},"_tags":["story","author_rjurney","story_22369482"],"author":"rjurney","created_at":"2020-02-19T20:06:38Z","created_at_i":1582142798,"num_comments":0,"objectID":"22369482","points":5,"story_id":22369482,"title":"Hand labeling is the past. The future is NoLabel AI with Snorkel","updated_at":"2024-09-20T05:43:39Z","url":"https://www.kdnuggets.com/2020/02/hand-labeling-past-future-nolabel-ai.html"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Someone"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"A re-fitted snorkel mask can save lives"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://www.air-wave.org/"}},"_tags":["story","author_Someone","story_23105262"],"author":"Someone","created_at":"2020-05-07T17:00:35Z","created_at_i":1588870835,"num_comments":0,"objectID":"23105262","points":1,"story_id":23105262,"title":"A re-fitted snorkel mask can save lives","updated_at":"2024-09-20T06:08:13Z","url":"https://www.air-wave.org/"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snorkel"},"story_text":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":""Google did the same in 2015 with another AI technology, TensorFlow, which lets anyone link computers into a neural network that can process data in a way analogous to our own biological brains."
sigh
http://www.cnet.com/news/google-offers-parsey-mcparseface-and-syntaxnet-ai-software-for-free/"},"title":{"matchLevel":"none","matchedWords":[],"value":"And the Wrongest Description of TensorFlow Goes To"}},"_tags":["story","author_snorkel","story_11693168","ask_hn"],"author":"snorkel","created_at":"2016-05-13T20:52:23Z","created_at_i":1463172743,"num_comments":0,"objectID":"11693168","points":10,"story_id":11693168,"story_text":""Google did the same in 2015 with another AI technology, TensorFlow, which lets anyone link computers into a neural network that can process data in a way analogous to our own biological brains."
sigh
http://www.cnet.com/news/google-offers-parsey-mcparseface-and-syntaxnet-ai-software-for-free/","title":"And the Wrongest Description of TensorFlow Goes To","updated_at":"2024-09-19T23:15:19Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ArtWomb"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"Sorel-20M: A Large Scale Benchmark Dataset for Malicious PE Detection"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://github.com/sophos-ai/SOREL-20M"}},"_tags":["story","author_ArtWomb","story_25430134"],"author":"ArtWomb","created_at":"2020-12-15T13:36:34Z","created_at_i":1608039394,"num_comments":0,"objectID":"25430134","points":1,"story_id":25430134,"title":"Sorel-20M: A Large Scale Benchmark Dataset for Malicious PE Detection","updated_at":"2024-09-20T07:27:49Z","url":"https://github.com/sophos-ai/SOREL-20M"},{"_highlightResult":{"author":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snork_alt"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"An open source AI tool to animate children's drawings"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/facebookresearch/AnimatedDrawings"}},"_tags":["story","author_snork_alt","story_35565681"],"author":"snork_alt","children":[35568886,35568891,35569060,35570122,35573141,35594848],"created_at":"2023-04-14T04:59:29Z","created_at_i":1681448369,"num_comments":14,"objectID":"35565681","points":318,"story_id":35565681,"title":"An open source AI tool to animate children's drawings","updated_at":"2024-09-20T13:52:04Z","url":"https://github.com/facebookresearch/AnimatedDrawings"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"AI writing has a homogeneity problem"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://usenoren.ai/blog/why-ai-writing-sounds-the-same"}},"_tags":["story","author_snoren","story_47367553"],"author":"snoren","children":[47367589,47367876,47368289],"created_at":"2026-03-13T18:00:26Z","created_at_i":1773424826,"num_comments":6,"objectID":"47367553","points":5,"story_id":47367553,"title":"AI writing has a homogeneity problem","updated_at":"2026-03-14T08:31:12Z","url":"https://usenoren.ai/blog/why-ai-writing-sounds-the-same"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Google says generative AI visibility is still SEO"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://developers.google.com/search/docs/fundamentals/ai-optimization-guide"}},"_tags":["story","author_snoren","story_48154387"],"author":"snoren","children":[48154660],"created_at":"2026-05-15T21:52:09Z","created_at_i":1778881929,"num_comments":0,"objectID":"48154387","points":5,"story_id":48154387,"title":"Google says generative AI visibility is still SEO","updated_at":"2026-05-16T08:36:01Z","url":"https://developers.google.com/search/docs/fundamentals/ai-optimization-guide"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"How to train AI on your writing style"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://usenoren.ai/blog/how-to-train-ai-on-your-writing-style"}},"_tags":["story","author_snoren","story_47479682"],"author":"snoren","created_at":"2026-03-22T17:11:15Z","created_at_i":1774199475,"num_comments":0,"objectID":"47479682","points":4,"story_id":47479682,"title":"How to train AI on your writing style","updated_at":"2026-03-22T22:39:34Z","url":"https://usenoren.ai/blog/how-to-train-ai-on-your-writing-style"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Two Brothers Are Trying to Put the Person Back into AI"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://www.wearefounders.uk/how-two-brothers-are-trying-to-put-the-person-back-into-ai/"}},"_tags":["story","author_snoren","story_48444735"],"author":"snoren","children":[48445431],"created_at":"2026-06-08T12:52:19Z","created_at_i":1780923139,"num_comments":0,"objectID":"48444735","points":3,"story_id":48444735,"title":"Two Brothers Are Trying to Put the Person Back into AI","updated_at":"2026-06-09T13:23:35Z","url":"https://www.wearefounders.uk/how-two-brothers-are-trying-to-put-the-person-back-into-ai/"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"What changes when AI reads you first"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://onomeokajevo.substack.com/p/stop-telling-ai-to-sound-like-you"}},"_tags":["story","author_snoren","story_48191180"],"author":"snoren","created_at":"2026-05-19T09:31:04Z","created_at_i":1779183064,"num_comments":0,"objectID":"48191180","points":3,"story_id":48191180,"title":"What changes when AI reads you first","updated_at":"2026-05-19T10:49:43Z","url":"https://onomeokajevo.substack.com/p/stop-telling-ai-to-sound-like-you"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"AI Is Computation"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://www.gabrielpickard.com/posts/ai-is-computation/"}},"_tags":["story","author_snoren","story_48388219"],"author":"snoren","children":[48388661],"created_at":"2026-06-03T18:55:35Z","created_at_i":1780512935,"num_comments":3,"objectID":"48388219","points":2,"story_id":48388219,"title":"AI Is Computation","updated_at":"2026-06-04T01:09:29Z","url":"https://www.gabrielpickard.com/posts/ai-is-computation/"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"smokel"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"AI tools for software engineers, but without the hype \u2013 with Simon Willison [video]"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.youtube.com/watch?v=uRuLgar5XZw"}},"_tags":["story","author_smokel","story_41649505"],"author":"smokel","created_at":"2024-09-25T17:05:22Z","created_at_i":1727283922,"num_comments":0,"objectID":"41649505","points":2,"story_id":41649505,"title":"AI tools for software engineers, but without the hype \u2013 with Simon Willison [video]","updated_at":"2024-09-25T17:24:05Z","url":"https://www.youtube.com/watch?v=uRuLgar5XZw"},{"_highlightResult":{"author":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snork_alt"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Realistic AI Based Text2speech"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://suno-ai.notion.site/Bark-Examples-5edae8b02a604b54a42244ba45ebc2e2"}},"_tags":["story","author_snork_alt","story_35695737"],"author":"snork_alt","created_at":"2023-04-25T02:04:03Z","created_at_i":1682388243,"num_comments":0,"objectID":"35695737","points":2,"story_id":35695737,"title":"Realistic AI Based Text2speech","updated_at":"2024-09-20T13:55:15Z","url":"https://suno-ai.notion.site/Bark-Examples-5edae8b02a604b54a42244ba45ebc2e2"},{"_highlightResult":{"author":{"fullyHighlighted":true,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snoren"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Why your AI writing sounds like everyone else's"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://usenoren.ai/blog/why-ai-writing-sounds-the-same"}},"_tags":["story","author_snoren","story_47349484"],"author":"snoren","children":[47349777],"created_at":"2026-03-12T12:10:07Z","created_at_i":1773317407,"num_comments":1,"objectID":"47349484","points":1,"story_id":47349484,"title":"Why your AI writing sounds like everyone else's","updated_at":"2026-03-12T14:03:35Z","url":"https://usenoren.ai/blog/why-ai-writing-sounds-the-same"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rishabhd"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"Sophos-ReversingLabs (SOREL) 20M sample malware dataset"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"https://ai.sophos.com/2020/12/14/sophos-reversinglabs-sorel-20-million-sample-malware-dataset/"}},"_tags":["story","author_rishabhd","story_25426534"],"author":"rishabhd","created_at":"2020-12-15T03:13:00Z","created_at_i":1608001980,"num_comments":0,"objectID":"25426534","points":1,"story_id":25426534,"title":"Sophos-ReversingLabs (SOREL) 20M sample malware dataset","updated_at":"2024-09-20T07:27:23Z","url":"https://ai.sophos.com/2020/12/14/sophos-reversinglabs-sorel-20-million-sample-malware-dataset/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"gpmcadam"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["snorkel","ai"],"value":"Over the years as "plain HTML"[1] websites in the mainstream have moved aside for JS-rich "Web Apps" I've started to appreciate the few sites that still exist as primarily HTML and CSS.
They load fast, respond instantly, and don't overwhelm you with newsletters, cookies, privacy or other pop-ups; but most importantly they are reliable and for whatever reason -- perhaps the Boring\u2122 hosting over popular cloud hosting providers -- they are always up.
Anyway, some of mine are listed below but I'd love to see if the HN community has some treasured sites they still love and use regularly (other than HN of course).
- EpGuides[2] - plain-text list of episodes for TV shows including original air date and links to episodes as CSV, Wikipedia, IMDB, etc.
- Snooker.org[3] - Lite snooker live scores, TV schedule, tournament brackets + API access
- Live Football on TV[4] - Tabular list of football (soccer) fixtures and which TV network is showing them (UK only)
--
[1] I'm aware that not all of them are plain HTML, they are likely rendered from various server-side backends and some even have ads but I'm using that term to mean essentially once it renders, it's static and has minimal if any client-side interaction
[2] e.g. https://epguides.com/Lost/
[3] https://www.snooker.org/
[4] https://www.live-footballontv.com/"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Favourite low-tech/plain HTML websites?"}},"_tags":["story","author_gpmcadam","story_38554515","ask_hn"],"author":"gpmcadam","children":[38555761,38559977],"created_at":"2023-12-07T09:25:49Z","created_at_i":1701941149,"num_comments":2,"objectID":"38554515","points":3,"story_id":38554515,"story_text":"Over the years as "plain HTML"[1] websites in the mainstream have moved aside for JS-rich "Web Apps" I've started to appreciate the few sites that still exist as primarily HTML and CSS.
They load fast, respond instantly, and don't overwhelm you with newsletters, cookies, privacy or other pop-ups; but most importantly they are reliable and for whatever reason -- perhaps the Boring\u2122 hosting over popular cloud hosting providers -- they are always up.
Anyway, some of mine are listed below but I'd love to see if the HN community has some treasured sites they still love and use regularly (other than HN of course).
- EpGuides[2] - plain-text list of episodes for TV shows including original air date and links to episodes as CSV, Wikipedia, IMDB, etc.
- Snooker.org[3] - Lite snooker live scores, TV schedule, tournament brackets + API access
- Live Football on TV[4] - Tabular list of football (soccer) fixtures and which TV network is showing them (UK only)
--
[1] I'm aware that not all of them are plain HTML, they are likely rendered from various server-side backends and some even have ads but I'm using that term to mean essentially once it renders, it's static and has minimal if any client-side interaction
[2] e.g. https://epguides.com/Lost/
[3] https://www.snooker.org/
[4] https://www.live-footballontv.com/","title":"Ask HN: Favourite low-tech/plain HTML websites?","updated_at":"2024-09-20T15:51:15Z"},{"_highlightResult":{"author":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snork_alt"},"title":{"matchLevel":"none","matchedWords":[],"value":"Robots that learn from videos of human activities and simulated interactions"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://ai.facebook.com/blog/robots-learning-video-simulation-artificial-visual-cortex-vc-1/"}},"_tags":["story","author_snork_alt","story_35421061"],"author":"snork_alt","created_at":"2023-04-03T10:14:47Z","created_at_i":1680516887,"num_comments":0,"objectID":"35421061","points":3,"story_id":35421061,"title":"Robots that learn from videos of human activities and simulated interactions","updated_at":"2024-09-20T13:45:26Z","url":"https://ai.facebook.com/blog/robots-learning-video-simulation-artificial-visual-cortex-vc-1/"},{"_highlightResult":{"author":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snork_alt"},"title":{"matchLevel":"none","matchedWords":[],"value":"Dozer: Fast API Data Serving from Multiple Sources with Real-Time Transformation"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://getdozer.io/blog/like-airbyte-elasticsearch-15x-faster/"}},"_tags":["story","author_snork_alt","story_36248454"],"author":"snork_alt","children":[36253073],"created_at":"2023-06-08T20:28:51Z","created_at_i":1686256131,"num_comments":1,"objectID":"36248454","points":2,"story_id":36248454,"title":"Dozer: Fast API Data Serving from Multiple Sources with Real-Time Transformation","updated_at":"2024-09-20T14:14:49Z","url":"https://getdozer.io/blog/like-airbyte-elasticsearch-15x-faster/"},{"_highlightResult":{"author":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"snork_alt"},"title":{"matchLevel":"none","matchedWords":[],"value":"Will Rust Takeover Data Engineering?"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://airbyte.com/blog/rust-for-data-engineering"}},"_tags":["story","author_snork_alt","story_35368692"],"author":"snork_alt","children":[35368739],"created_at":"2023-03-30T06:07:06Z","created_at_i":1680156426,"num_comments":1,"objectID":"35368692","points":2,"story_id":35368692,"title":"Will Rust Takeover Data Engineering?","updated_at":"2024-09-20T13:40:50Z","url":"https://airbyte.com/blog/rust-for-data-engineering"},{"_highlightResult":{"author":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["snorkel"],"value":"sporkl_l"},"title":{"matchLevel":"none","matchedWords":[],"value":"Some Basic LLM Etiquette"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://steenbok.space/blog/ai-etiquette/"}},"_tags":["story","author_sporkl_l","story_48769662"],"author":"sporkl_l","created_at":"2026-07-03T01:39:37Z","created_at_i":1783042777,"num_comments":0,"objectID":"48769662","points":2,"story_id":48769662,"title":"Some Basic LLM Etiquette","updated_at":"2026-07-03T03:11:51Z","url":"https://steenbok.space/blog/ai-etiquette/"}],"hitsPerPage":50,"nbHits":42,"nbPages":1,"page":0,"params":"query=Snorkel+AI&tags=story&hitsPerPage=50&advancedSyntax=true&analyticsTags=backend","processingTimeMS":9,"processingTimingsMS":{"_request":{"queue":1,"roundTrip":17},"afterFetch":{"format":{"highlighting":1,"total":1}},"fetch":{"query":6,"scanning":1,"total":8},"total":9},"query":"Snorkel AI","serverTimeMS":12}