[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EMCPPFFEG1DFJ0RTEYGN3P":3,"skill-same:01M7EMCPPFFEG1DFJ0RTEYGN3P":47},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7EMCPPFFEG1DFJ0RTEYGN3P","baseten-guide",null,"Deploy ML models to production with Baseten — high-performance inference with autoscaling and custom runtimes.","TanNhatCMS\u002Fjarvis","TanNhatCMS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F113796420?v=4","01M753MV17KG5CD3152G3WQEQY","KietNT","01M7EKEDYVYQ6PWVJJMZ911VMX",0,"MIT","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fams-baseten-guide\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: baseten-guide\ndescription: Deploy ML models to production with Baseten — high-performance inference with autoscaling and custom runtimes.\ncategory: ai-research\n---\n\n## Overview\n\nBaseten is a model-deployment platform focused on production inference\nperformance: deploy open models (or your own) behind autoscaled, low-latency APIs\nwith strong performance engineering (custom inference engines, optimized\nserving stacks). The pitch is \"self-hosted performance without self-hosting\" —\nyou get serious inference optimization managed for you, with dedicated capacity\noptions for strict SLOs.\n\nWhere serverless GPU platforms optimize for convenience and scale-to-zero,\nBaseten optimizes for the production middle: sustained traffic, latency SLOs,\nand cost-per-token at volume. If your workload has graduated from \"run this\nmodel sometimes\" to \"serve this model reliably at scale,\" Baseten is built for\nthat transition.\n\nThe evaluation lens: measure p99 latency and cost per million tokens at your\nactual traffic shape. Baseten's value shows up in those numbers, not in feature\nlists.\n\n## When to use\n\n- Production LLM serving with latency SLOs (p99 matters, not just averages).\n- High-volume open-model inference where cost-per-token at scale is the metric.\n- Deploying fine-tuned models behind production-grade endpoints.\n- Multi-modal model serving (the platform covers more than text).\n- Teams that need inference performance engineering without hiring for it.\n- Dedicated capacity for isolation and predictable performance.\n\n## Core concepts\n\n- **Model deployments**: versioned deployments of models behind API endpoints.\n  Deploy from popular open models or your own weights\u002Ffine-tunes. Versioning\n  lets you roll forward and back safely.\n- **Performance-optimized serving**: the platform applies inference\n  optimizations (efficient attention kernels, batching, quantization options) so\n  you don't have to. Understand what's applied to your deployment — it affects\n  both latency and output behavior.\n- **Autoscaling**: capacity scales with traffic within configured bounds.\n  Configure min\u002Fmax replicas deliberately: min for baseline latency (warm\n  capacity), max for cost protection.\n- **Dedicated deployments**: reserved GPUs for your workload — no noisy\n  neighbors, predictable latency. The right choice when shared capacity variance\n  violates SLOs.\n- **Custom models and fine-tunes**: bring your own weights or train on the\n  platform; serve the result on the same optimized stack. Keeps the\n  customize→serve loop tight.\n- **Development vs. production tiers**: lighter\u002Fcheaper options for dev and\n  staging, performance tiers for production. Don't benchmark dev-tier latency and\n  extrapolate to production.\n- **Observability**: request logging, latency metrics, and GPU utilization.\n  Wire these into your monitoring — inference issues show up here before users\n  report them.\n- **Cost model**: pay for the capacity you reserve plus usage. Model your\n  expected cost at baseline and peak traffic before committing.\n\n## Practical workflow\n\n1. **Define SLOs first.** Target p50\u002Fp99 latency, throughput, and availability.\n   Everything — deployment tier, replica counts, dedicated vs. shared — follows\n   from these numbers.\n2. **Deploy and benchmark.** Deploy the model, then load-test at realistic\n   traffic shape (not just steady-state — include bursts). Measure p99 latency\n   and tokens\u002Fsec; tune replica bounds and instance types.\n3. **Validate output quality.** Optimized serving stacks (quantization, custom\n   kernels) can subtly change outputs. Run your eval set against the deployed\n   endpoint, not just the reference model.\n4. **Configure autoscaling deliberately.** Set min replicas for warm baseline\n   capacity (cold scaling hurts p99), max replicas as a cost circuit breaker.\n   Test scale-up behavior under a traffic spike.\n5. **Set up staging.** Mirror production config in staging for safe rollout\n   testing. Canary new model versions through staging before production.\n6. **Version and roll back safely.** Deploy new versions alongside old; shift\n   traffic gradually; keep the previous version deployable for instant rollback.\n7. **Monitor continuously.** Track latency percentiles, error rates, GPU\n   utilization, and cost. Alert on SLO breaches and cost anomalies.\n\nChecklist for a Baseten production deployment:\n- SLOs defined; load test passed at realistic traffic shape.\n- Output quality validated on the deployed (optimized) stack.\n- Autoscaling min\u002Fmax set and spike-tested.\n- Staging mirrors production; canary process defined.\n- Latency, error, utilization, and cost monitoring in place.\n\n## Common pitfalls\n\n- **Benchmarking the wrong tier.** Dev-tier numbers don't predict production\n  latency. Benchmark the tier you'll actually run.\n- **Skipping output validation.** Assuming the optimized stack is\n  behavior-identical to the reference model. Quantization and custom kernels can\n  shift outputs — verify.\n- **Min replicas at zero for latency-sensitive APIs.** Scale-from-zero saves\n  money and destroys p99. Warm baseline capacity is the cost of latency SLOs.\n- **No max-replica cost guard.** Autoscaling without an upper bound during a\n  traffic anomaly or retry storm. Set the max; alert approaching it.\n- **Traffic shape mismatch.** Load-testing steady state when production has\n  bursts. Test bursts, cold traffic patterns, and sustained peaks.\n- **Big-bang version upgrades.** Swapping model versions without canarying.\n  Inference behavior changes are user-facing — roll out gradually.\n- **Ignoring GPU utilization.** Low utilization means you're overprovisioned\n  (wasting money); saturated means no headroom (latency risk). Right-size from\n  the metrics.\n- **No rollback plan.** A bad deploy with no quick revert path. Keep the previous\n  version warm and the rollback procedure rehearsed.\n","## Overview\n\nBaseten is a model-deployment platform focused on production inference\nperformance: deploy open models (or your own) behind autoscaled, low-latency APIs\nwith strong performance engineering (custom inference engines, optimized\nserving stacks). The pitch is \"self-hosted performance without self-hosting\" —\nyou get serious inference optimization managed for you, with dedicated capacity\noptions for strict SLOs.\n\nWhere serverless GPU platforms optimize for convenience and scale-to-zero,\nBaseten optimizes for the production middle: sustained traffic, latency SLOs,\nand cost-per-token at volume. If your workload has graduated from \"run this\nmodel sometimes\" to \"serve this model reliably at scale,\" Baseten is built for\nthat transition.\n\nThe evaluation lens: measure p99 latency and cost per million tokens at your\nactual traffic shape. Baseten's value shows up in those numbers, not in feature\nlists.\n\n## When to use\n\n- Production LLM serving with latency SLOs (p99 matters, not just averages).\n- High-volume open-model inference where cost-per-token at scale is the metric.\n- Deploying fine-tuned models behind production-grade endpoints.\n- Multi-modal model serving (the platform covers more than text).\n- Teams that need inference performance engineering without hiring for it.\n- Dedicated capacity for isolation and predictable performance.\n\n## Core concepts\n\n- **Model deployments**: versioned deployments of models behind API endpoints.\n  Deploy from popular open models or your own weights\u002Ffine-tunes. Versioning\n  lets you roll forward and back safely.\n- **Performance-optimized serving**: the platform applies inference\n  optimizations (efficient attention kernels, batching, quantization options) so\n  you don't have to. Understand what's applied to your deployment — it affects\n  both latency and output behavior.\n- **Autoscaling**: capacity scales with traffic within configured bounds.\n  Configure min\u002Fmax replicas deliberately: min for baseline latency (warm\n  capacity), max for cost protection.\n- **Dedicated deployments**: reserved GPUs for your workload — no noisy\n  neighbors, predictable latency. The right choice when shared capacity variance\n  violates SLOs.\n- **Custom models and fine-tunes**: bring your own weights or train on the\n  platform; serve the result on the same optimized stack. Keeps the\n  customize→serve loop tight.\n- **Development vs. production tiers**: lighter\u002Fcheaper options for dev and\n  staging, performance tiers for production. Don't benchmark dev-tier latency and\n  extrapolate to production.\n- **Observability**: request logging, latency metrics, and GPU utilization.\n  Wire these into your monitoring — inference issues show up here before users\n  report them.\n- **Cost model**: pay for the capacity you reserve plus usage. Model your\n  expected cost at baseline and peak traffic before committing.\n\n## Practical workflow\n\n1. **Define SLOs first.** Target p50\u002Fp99 latency, throughput, and availability.\n   Everything — deployment tier, replica counts, dedicated vs. shared — follows\n   from these numbers.\n2. **Deploy and benchmark.** Deploy the model, then load-test at realistic\n   traffic shape (not just steady-state — include bursts). Measure p99 latency\n   and tokens\u002Fsec; tune replica bounds and instance types.\n3. **Validate output quality.** Optimized serving stacks (quantization, custom\n   kernels) can subtly change outputs. Run your eval set against the deployed\n   endpoint, not just the reference model.\n4. **Configure autoscaling deliberately.** Set min replicas for warm baseline\n   capacity (cold scaling hurts p99), max replicas as a cost circuit breaker.\n   Test scale-up behavior under a traffic spike.\n5. **Set up staging.** Mirror production config in staging for safe rollout\n   testing. Canary new model versions through staging before production.\n6. **Version and roll back safely.** Deploy new versions alongside old; shift\n   traffic gradually; keep the previous version deployable for instant rollback.\n7. **Monitor continuously.** Track latency percentiles, error rates, GPU\n   utilization, and cost. Alert on SLO breaches and cost anomalies.\n\nChecklist for a Baseten production deployment:\n- SLOs defined; load test passed at realistic traffic shape.\n- Output quality validated on the deployed (optimized) stack.\n- Autoscaling min\u002Fmax set and spike-tested.\n- Staging mirrors production; canary process defined.\n- Latency, error, utilization, and cost monitoring in place.\n\n## Common pitfalls\n\n- **Benchmarking the wrong tier.** Dev-tier numbers don't predict production\n  latency. Benchmark the tier you'll actually run.\n- **Skipping output validation.** Assuming the optimized stack is\n  behavior-identical to the reference model. Quantization and custom kernels can\n  shift outputs — verify.\n- **Min replicas at zero for latency-sensitive APIs.** Scale-from-zero saves\n  money and destroys p99. Warm baseline capacity is the cost of latency SLOs.\n- **No max-replica cost guard.** Autoscaling without an upper bound during a\n  traffic anomaly or retry storm. Set the max; alert approaching it.\n- **Traffic shape mismatch.** Load-testing steady state when production has\n  bursts. Test bursts, cold traffic patterns, and sustained peaks.\n- **Big-bang version upgrades.** Swapping model versions without canarying.\n  Inference behavior changes are user-facing — roll out gradually.\n- **Ignoring GPU utilization.** Low utilization means you're overprovisioned\n  (wasting money); saturated means no headroom (latency risk). Right-size from\n  the metrics.\n- **No rollback plan.** A bad deploy with no quick revert path. Keep the previous\n  version warm and the rollback procedure rehearsed.\n",{"name":5,"category":24,"description":7},"ai-research","main","af388a77c2fe5e0b4fd879cca4300a0ffa9a6579","skills\u002Fams-baseten-guide",1,"https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis","https:\u002F\u002Fgithub.com\u002FTanNhatCMS","https:\u002F\u002Fmrkiet.name.vn","2026-11-07 14:23:08",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add TanNhatCMS\u002Fjarvis --skill baseten-guide",{"tool":39,"command":40},"bunx","bunx skills add TanNhatCMS\u002Fjarvis --skill baseten-guide",{"tool":42,"command":43},"pnpm","pnpm dlx skills add TanNhatCMS\u002Fjarvis --skill baseten-guide","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fams-baseten-guide\u002FSKILL.md",true,[],[48,55,63,70,77,84],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7EWYJR0J8DARY7E59Z6Y9MV","web-scrape","Turn web pages into LLM-ready markdown via the Firecrawl API (renders JS, anti-bot handling). Use when browser_open\u002Ffetch fails on JS-heavy pages, or when you need clean markdown\u002Fstructured extraction from a site. Requires FIRECRAWL_API_KEY.","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fweb-scrape\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":59,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7EWYBN5P6F2AJG3FR24T9BP","design-taste-frontend","tasteskill v2, the anti-slop frontend skill. Art direction for websites, distilled from a study of 247 curated gallery sites (One Page Love, Awwwards Sites of the Day). Deals researched directions (idea, composition, type, palette, photographs, section blueprints) from a library, enforces a strict no-slop house style with a real-browser inspector, and runs a look-and-fix loop on screenshots. Use this whenever you build, design, redesign, restyle, polish or review a landing page, marketing site, portfolio, product page, one-pager, hero or any public-facing web page, in any stack (HTML\u002FCSS, React, Next.js, Vue, Svelte, Astro, Tailwind) — also when the user says a page looks generic, templated, \"AI-made\", sloppy, or asks to make it look designed, premium, award-level or less like every other site. Not for dashboards or app screens.","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Ftaste-design\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Ftaste-design\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7EWY8FQFNQ2GB7NZZWGSSDX","test-driven-development","Use when implementing any feature or bugfix, before writing implementation code","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-tdd\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7EWY5SHBFRH1HDHVWM9K7TV","systematic-debugging","Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-debugging\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7EWY3NM0Z0JAKS7HDZ24K66","requesting-code-review","Use when completing tasks, implementing major features, or before merging to verify work meets requirements","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-code-review\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7EWY0WRMWX4M9ZFY0WKC593","brainstorming","You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).","https:\u002F\u002Fgithub.com\u002FTanNhatCMS\u002Fjarvis\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers-brainstorming\u002FSKILL.md",[],[]]