{"slug": "ray", "name": "Ray", "category": "dev-libs", "category_name": "Libraries and frameworks", "description": "", "recommenders": 5, "commenters": 12, "warners": 2, "score": 3.244, "rising": -1.64, "first": "2025-11-06", "latest": "2026-10-02", "why": "We wrote up the full Ray architecture we use in production on KubeRay/GKE.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 5, "neutral": 4, "compare": 1, "warn": 2}, "spark": [1, 1, 0, 2, 0, 0, 0, 0, 1, 0, 0, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "We wrote up the full Ray architecture we use in production on KubeRay/GKE.", "kind": "", "icon": true, "domain": "ray.io", "github_repo": null, "official_url": "https://www.ray.io", "pricing": "", "aliases": ["Ray"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 782299234, "item_id": 48893702, "url": "https://news.ycombinator.com/item?id=48893702", "by": "skreem", "date": "2026-07-13", "stance": "recommend", "quote": "stitched together with Ray autoscaling nodes it can get really nice", "excerpt": "Oh this is awesome, will take a look. I\u2019ve had rough experiences with Kubeflow in the past, recently I\u2019ve been using Coder workspaces https://coder.com/ and it\u2019s pretty good but less ML/AI integrated (but, stitched together with Ray autoscaling nodes it can get really nice)", "context": "", "use_case": "autoscaling compute", "switched_from": "", "story": "Ask HN: What Are You Working On? (July 2026)"}, {"id": 754451648, "item_id": 47153228, "url": "https://news.ycombinator.com/item?id=47153228", "by": "Beefin", "date": "2026-02-25", "stance": "recommend", "quote": "We wrote up the full Ray architecture we use in production on KubeRay/GKE.", "excerpt": "We process video, images, and documents through 20+ ML models simultaneously at Mixpeek. A single 10-minute video triggers transcription, visual embeddings, scene descriptions, face detection, object detection, brand safety classification, and more \u2014 all in parallel with different compute requirements. We wrote up the full Ray architecture we use in production on KubeRay/GKE. Not a tutorial \u2014 more of a \"here's what w", "context": "works at Mixpeek", "use_case": "parallel ML pipelines", "switched_from": "", "story": "We run 20M models in parallel on Ray"}, {"id": 749759842, "item_id": 46859990, "url": "https://news.ycombinator.com/item?id=46859990", "by": "smithclay", "date": "2026-02-02", "stance": "recommend", "quote": "have also noticed this pattern in Modal [1] and Ray [2] and all for it.", "excerpt": "This looks cool, love the approach of a more \"pythonic\" infra as code -- have also noticed this pattern in Modal [1] and Ray [2] and all for it. [1] https://modal.com/docs/examples/hackernews_alerts [2] https://docs.ray.io/en/latest/ray-overview/getting-started.html edit: forgot to mention StackQL [3], which is somewhat similar but takes a SQL-first vs python-first approach. The IaC space is about to get a lot more i", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Stelvio \u2013 Ship Python to AWS"}, {"id": 737978112, "item_id": 46123632, "url": "https://news.ycombinator.com/item?id=46123632", "by": "northlondoner", "date": "2025-12-02", "stance": "recommend", "quote": "One thing valuable is \"Fault Tolerance\"", "excerpt": "One thing valuable is \"Fault Tolerance\", https://docs.ray.io/en/latest/ray-core/fault-tolerance.html", "context": "", "use_case": "", "switched_from": "", "story": "Ray Keynote: de-facto for compositional learning and a distributed heterogeneous [video]"}, {"id": 733468256, "item_id": 45841766, "url": "https://news.ycombinator.com/item?id=45841766", "by": "zhethoughts", "date": "2025-11-06", "stance": "recommend", "quote": "https://docs.ray.io/en/latest/", "excerpt": "https://docs.ray.io/en/latest/", "context": "", "use_case": "parallel computing", "switched_from": "", "story": "I want a good parallel language [video]"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 749264128, "item_id": 46829008, "url": "https://news.ycombinator.com/item?id=46829008", "by": "thenaturalist", "date": "2026-01-30", "stance": "warn", "quote": "The Ray team even rejected Lambda support expressedly as far back as 2020", "excerpt": "> \"Forget about managing complex server infrastructure for your database needs.\" So what does this run on then? No docs, it's not possible to find any deployment guides for Ray using serverless solutions like Lambda, Cloud Functions or be it your own Firecracker. Instead, every other post mentions EKS or EC2. The Ray team even rejected Lambda support expressedly as far back as 2020 [0]. Uuuuuugh. No thanks! shiver I'", "context": "", "use_case": "", "switched_from": "", "story": "Quack-Cluster: A Serverless Distributed SQL Query Engine with DuckDB and Ray"}, {"id": 734822725, "item_id": 45926420, "url": "https://news.ycombinator.com/item?id=45926420", "by": "faizshah", "date": "2025-11-14", "stance": "warn", "quote": "Dask and Ray were also way too slow", "excerpt": "I had to do something like this for a few TB of json recently. The unique thing about this workload was it was a ton of small 10-20mb files. I found that clickhouse was the fastest, but duckdb was the simplest to work with it usually just works. DuckDB was close enough to the max performance from clickhouse. I tried flink & pyspark but they were way slower (like 3-5x) than clickhouse and the code was kind of annoying", "context": "", "use_case": "big data processing", "switched_from": "", "story": "650GB of Data (Delta Lake on S3). Polars vs. DuckDB vs. Daft vs. Spark"}]}], "who": [{"by": "Beefin", "context": "works at Mixpeek", "url": "https://news.ycombinator.com/item?id=47153228", "date": "2026-02-25"}, {"by": "mycelia", "context": "Ray developer", "url": "https://news.ycombinator.com/item?id=46628831", "date": "2026-01-14"}, {"by": "Grosvenor", "context": "data scientist with 20+ years experience", "url": "https://news.ycombinator.com/item?id=46110952", "date": "2025-12-01"}], "tags": [], "use_cases": ["autoscaling compute", "personal finance tracking", "parallel ml pipelines", "distributed compute orchestration", "distributed computing", "data science consulting"], "compared": [], "makers": [{"id": 745178352, "item_id": 46573647, "url": "https://news.ycombinator.com/item?id=46573647", "by": "techspecs", "date": "2026-01-11", "stance": "recommend", "quote": "I have a temporary install available while the Chrome Web Store listing is pending", "excerpt": "It\u2019s not open source at the moment. I initially explored a fully client-side approach (including WebAssembly), but it didn\u2019t work well in practice. Real-time audio transcription and multi-language translation are both compute-intensive, and browser-only solutions ran into performance and reliability limits, especially for longer videos and live streams. Using a dedicated backend allowed more consistent latency and ac", "context": "", "use_case": "YouTube live subtitles", "switched_from": "", "story": "Generating real-time subtitles from YouTube audio (even when captions fail)"}], "threads": [{"url": "https://news.ycombinator.com/item?id=46123572", "title": "Ray Keynote: de-facto for compositional learning and a distributed heterogeneous [video]", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49922569", "title": "Ask HN: Who is hiring? (October 2026)", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48884984", "title": "Ask HN: What Are You Working On? (July 2026)", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48090276", "title": "item 48090276", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=47153227", "title": "We run 20M models in parallel on Ray", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=46856899", "title": "Show HN: Stelvio \u2013 Ship Python to AWS", "mentions": 1}], "sentiment_by_month": {"2025-11": {"recommend": 1, "compare": 1, "warn": 1}, "2025-12": {"neutral": 1, "recommend": 1}, "2026-01": {"neutral": 1, "warn": 1}, "2026-02": {"recommend": 2}, "2026-03": {}, "2026-04": {}, "2026-05": {"neutral": 1}, "2026-06": {}, "2026-07": {"recommend": 1}, "2026-08": {}, "2026-09": {}, "2026-10": {"neutral": 1}}, "quotes": [{"id": 798920401, "item_id": 49932525, "url": "https://news.ycombinator.com/item?id=49932525", "by": "l1am0", "date": "2026-10-02", "stance": "neutral", "quote": "cloud + HPC clusters, K8s, Ray, Dask", "excerpt": "Proxima Fusion | Senior Software Engineer (Dev Platform) | Munich, Germany | ONSITE | Full-time We're building the world's first commercial stellarator fusion power plant. First spin-out from the Max Planck Institute for Plasma Physics, based on Wendelstein 7-X (the largest stellarator ever built), \u20ac650M+ raised, teams in Munich, Zurich and Oxford. Finding a good stellarator design means searching an enormous high-di", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: Who is hiring? (October 2026)"}, {"id": 782299234, "item_id": 48893702, "url": "https://news.ycombinator.com/item?id=48893702", "by": "skreem", "date": "2026-07-13", "stance": "recommend", "quote": "stitched together with Ray autoscaling nodes it can get really nice", "excerpt": "Oh this is awesome, will take a look. I\u2019ve had rough experiences with Kubeflow in the past, recently I\u2019ve been using Coder workspaces https://coder.com/ and it\u2019s pretty good but less ML/AI integrated (but, stitched together with Ray autoscaling nodes it can get really nice)", "context": "", "use_case": "autoscaling compute", "switched_from": "", "story": "Ask HN: What Are You Working On? (July 2026)"}, {"id": 769547857, "item_id": 48096741, "url": "https://news.ycombinator.com/item?id=48096741", "by": "adamthegoalie", "date": "2026-05-11", "stance": "neutral", "quote": "I recently opened a PR against this AI personal finance tool Ray", "excerpt": "That is such a good point. I recently opened a PR against this AI personal finance tool Ray https://github.com/cdinnison/ray-finance/pull/8 to add an Apple Card import feature, since Apple Card is not supported by Plaid. I built the manual import feature, opened the PR, and then ran a code review. What I hadn't thought about when I built the feature, was the myriad ways that the implications of importing data from Ap", "context": "", "use_case": "Personal finance tracking", "switched_from": "", "story": null}, {"id": 754451648, "item_id": 47153228, "url": "https://news.ycombinator.com/item?id=47153228", "by": "Beefin", "date": "2026-02-25", "stance": "recommend", "quote": "We wrote up the full Ray architecture we use in production on KubeRay/GKE.", "excerpt": "We process video, images, and documents through 20+ ML models simultaneously at Mixpeek. A single 10-minute video triggers transcription, visual embeddings, scene descriptions, face detection, object detection, brand safety classification, and more \u2014 all in parallel with different compute requirements. We wrote up the full Ray architecture we use in production on KubeRay/GKE. Not a tutorial \u2014 more of a \"here's what w", "context": "works at Mixpeek", "use_case": "parallel ML pipelines", "switched_from": "", "story": "We run 20M models in parallel on Ray"}, {"id": 749759842, "item_id": 46859990, "url": "https://news.ycombinator.com/item?id=46859990", "by": "smithclay", "date": "2026-02-02", "stance": "recommend", "quote": "have also noticed this pattern in Modal [1] and Ray [2] and all for it.", "excerpt": "This looks cool, love the approach of a more \"pythonic\" infra as code -- have also noticed this pattern in Modal [1] and Ray [2] and all for it. [1] https://modal.com/docs/examples/hackernews_alerts [2] https://docs.ray.io/en/latest/ray-overview/getting-started.html edit: forgot to mention StackQL [3], which is somewhat similar but takes a SQL-first vs python-first approach. The IaC space is about to get a lot more i", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Stelvio \u2013 Ship Python to AWS"}, {"id": 749264128, "item_id": 46829008, "url": "https://news.ycombinator.com/item?id=46829008", "by": "thenaturalist", "date": "2026-01-30", "stance": "warn", "quote": "The Ray team even rejected Lambda support expressedly as far back as 2020", "excerpt": "> \"Forget about managing complex server infrastructure for your database needs.\" So what does this run on then? No docs, it's not possible to find any deployment guides for Ray using serverless solutions like Lambda, Cloud Functions or be it your own Firecracker. Instead, every other post mentions EKS or EC2. The Ray team even rejected Lambda support expressedly as far back as 2020 [0]. Uuuuuugh. No thanks! shiver I'", "context": "", "use_case": "", "switched_from": "", "story": "Quack-Cluster: A Serverless Distributed SQL Query Engine with DuckDB and Ray"}, {"id": 746061296, "item_id": 46628831, "url": "https://news.ycombinator.com/item?id=46628831", "by": "mycelia", "date": "2026-01-14", "stance": "neutral", "quote": "As a Ray developer", "excerpt": "Lot of cool stuff coming up! As a Ray developer, I focus more on the orchestration layer, so I'm excited about things like Elastic Expert Parallelism, posttraining enhancements like colocated trainer/engines, and deploying DSV4 (rumors are the architecture will be complex). vLLM roadmap is here for reference: http://roadmap.vllm.ai/", "context": "Ray developer", "use_case": "Distributed compute orchestration", "switched_from": "", "story": "VLLM Large Scale Serving: DeepSeek 2.2k Tok/S/H200 with Wide-EP"}, {"id": 737979776, "item_id": 46123736, "url": "https://news.ycombinator.com/item?id=46123736", "by": "northlondoner", "date": "2025-12-02", "stance": "neutral", "quote": "Ray over internet is also possible", "excerpt": "Ray over internet is also possible as well as K8 https://docs.ray.io/en/latest/cluster/kubernetes/index.html", "context": "", "use_case": "distributed computing", "switched_from": "", "story": "Ray Keynote: de-facto for compositional learning and a distributed heterogeneous [video]"}, {"id": 737978112, "item_id": 46123632, "url": "https://news.ycombinator.com/item?id=46123632", "by": "northlondoner", "date": "2025-12-02", "stance": "recommend", "quote": "One thing valuable is \"Fault Tolerance\"", "excerpt": "One thing valuable is \"Fault Tolerance\", https://docs.ray.io/en/latest/ray-core/fault-tolerance.html", "context": "", "use_case": "", "switched_from": "", "story": "Ray Keynote: de-facto for compositional learning and a distributed heterogeneous [video]"}, {"id": 737775234, "item_id": 46110952, "url": "https://news.ycombinator.com/item?id=46110952", "by": "Grosvenor", "date": "2025-12-01", "stance": "neutral", "quote": "I use the normal stacks you'd expect. Python, Pytorch, Spark/ray, Jupyter/Merimo, AWS, Postgres, Mathematica", "excerpt": ".-~~\\ / \\ _ ~x .-~_)_ ~x\".-~ ~-. _ ( / \\ _ || T o o Y || ==:l l < ! I;== \\\\ \\ .__/ / // \\\\ ,r\"-,___.-'r.// }^ \\.( ) _.'//. / }~Xi--~ // \\ Y Y I\\ \\ \" Y | | |o\\ \\ | | l_l Y T | l \"o l_j ! \\ / ___,.---^. o .^---.._____ \"~~~ \" ~ ~~~\" SEEKING WORK - Data scientist, consulting & fractional leadership, US/remote worldwide, email in profile. All I want for christmas is some gnarly problems to chew on, otherwise it's coal for", "context": "data scientist with 20+ years experience", "use_case": "data science consulting", "switched_from": "", "story": "Ask HN: Who wants to be hired? (December 2025)"}, {"id": 734822725, "item_id": 45926420, "url": "https://news.ycombinator.com/item?id=45926420", "by": "faizshah", "date": "2025-11-14", "stance": "warn", "quote": "Dask and Ray were also way too slow", "excerpt": "I had to do something like this for a few TB of json recently. The unique thing about this workload was it was a ton of small 10-20mb files. I found that clickhouse was the fastest, but duckdb was the simplest to work with it usually just works. DuckDB was close enough to the max performance from clickhouse. I tried flink & pyspark but they were way slower (like 3-5x) than clickhouse and the code was kind of annoying", "context": "", "use_case": "big data processing", "switched_from": "", "story": "650GB of Data (Delta Lake on S3). Polars vs. DuckDB vs. Daft vs. Spark"}, {"id": 734317650, "item_id": 45894853, "url": "https://news.ycombinator.com/item?id=45894853", "by": "cosmic_quanta", "date": "2025-11-11", "stance": "compare", "quote": "There's no reason the Haskell community can't have something like Dask or Ray", "excerpt": "This is great! I'm particularly excited by the prospect of bringing together the ecosystem of data science tools, with Cloud Haskell[0]. There's no reason the Haskell community can't have something like Dask or Ray, but with Haskell's stellar type system. [0]: https://github.com/haskell-distributed/distributed-process", "context": "", "use_case": "", "switched_from": "", "story": "Welcome to DataHaskell"}, {"id": 733468256, "item_id": 45841766, "url": "https://news.ycombinator.com/item?id=45841766", "by": "zhethoughts", "date": "2025-11-06", "stance": "recommend", "quote": "https://docs.ray.io/en/latest/", "excerpt": "https://docs.ray.io/en/latest/", "context": "", "use_case": "parallel computing", "switched_from": "", "story": "I want a good parallel language [video]"}]}