Machine learning

Build AI-powered browser agents that autonomously navigate, scrape, fill forms, and extract data from dynamic websites using LLM reasoning + Playwright.

Set up and optimize a complete local AI infrastructure — Ollama, Open WebUI, local RAG pipelines, and private model serving.

Builds AI-powered trading and financial analysis agents with multi-agent LLM architectures: research pipelines, sentiment signals, portfolio optimization…

Processar, comparar ou reindexar artigos acadêmicos no RAG MVLI com proveniência entre fonte, trecho e evidência.

Use when tasks need optional PyCaret model training, ML factor generation, inference wrappers, feature importance, or sparse LASSO weight generation.

Use when a mostly complete ML conference paper needs self-review, pre-submission QA, camera-ready checking, section-by-section critique, citation-risk…

Audit how a brand appears in AI-powered search (ChatGPT, Perplexity, Claude, Gemini).

Expert guidance on requesting, renewing, and revoking certificates in the Wendy ecosystem via pki-core.

Security invariants every Wendy interop change must respect, distilled from the AAA contract.

Show AI/LLM spending, budget usage, and operational cost metrics.

Find blog posts semantically similar to a topic — pulls from the posts embedding source.

Persistent PM knowledge base, competitive intel, customer notes, strategy, roadmap, AI market.

Choose OpenRouter models and routers, query live catalog data, use structured outputs and batch inference, and debug routing with metadata.

Query OpenTelemetry telemetry (traces, metrics, logs, LLM analytics) via the autotel CLI.

使用 alibaba/open-code-review（ocr CLI）对 Git 变更执行 AI 代码评审：OCR 以确定性工程完成文件选择与规则匹配，产出行级定位、按严重度分级的结构化意见（bug/安全/性能等）。默认委托模式由当前 Agent 执行评审，无需配置额外 LLM；也可切换 OCR…

Use when deploying, operating, debugging, or composing the Antioch Workbench integration, live OpenPI inference, and dataset-based policy training.

Use when Cosmos3 setup, fetch, inference, CUDA, uv, Docker, Hugging Face, GitHub, NGC, or checkpoint staging fails in NPA or in an upstream Cosmos framework…

Use when setting up Cosmos3 access through NPA, checking source or Hugging Face reachability, staging the public Cosmos3 framework and checkpoint cache, or…

Convex is the backend agents get right on the first try: an all-TypeScript reactive platform where the database, server functions, scheduling, file storage…

Suggest the matching Convex component when the user hand-rolls a pattern it already solves (crons, sharded-counter, rate-limiter, storage, search, presence…

Generate vector embeddings through DurinDoor using a model discovered from /v1/models/embedding.

Master LLM-as-a-Judge evaluation techniques including direct scoring, pairwise comparison, rubric generation, and bias mitigation.

Design and build LLM-powered projects from ideation through deployment.

Design multi-agent systems, agent loops, and eval harnesses.