Machine learning

Guided workflow for taking any ML codebase — including one with no experiment tracking at all, or one full of MLflow 2-era idioms — to a production-grade…

Production Ray (open-source, pinned to 2.57) for classic-ML workloads from training to serving — Ray Train, Tune, Data, Serve, Core, and cluster deployment on…

LLM workloads on open-source Ray (pinned to 2.57) — OpenAI-compatible serving with ray.serve.llm (vLLM-backed LLMConfig + build_openai_app) and batch…

用于诊断、评审或评分整份简历。当用户要求「诊断简历」「简历有什么问题」「看看简历哪里要改」「简历打分」「评审简历」「resume review」，或上传 PDF/DOCX 并要求分析问题、给出评分或修改建议时触发。输出 3 类诊断（NO_DATA / NO_FILL / DESCRIPTION_SUGGEST）和…

This skill should be used when designing, planning, implementing, or reviewing any system that involves LLM agents, tool use, prompt construction, or agentic…

Solve KCTF model backdoors and small linear/ReLU crackmes from raw weights.

Build, review, or harden AI agents, LLM apps, RAG, tool-calling, coding agents, and agentic workflows with secure-by-default controls.

Tool for preparing PDF figure files for embedding in LaTeX (e.g., with \includegraphics).

Compute and analyze embeddings for dataset quality, distribution comparison, semantic deduplication, diversity measurement, and similarity-based filtering.

Use locally-hosted LLMs (vLLM/SGLang) for dataset filtering, quality scoring, rewriting, labeling, and synthetic data generation.

Use this skill for web search, extraction, mapping, crawling, and research via Tavily’s REST API when web searches are needed and no built-in tool is…

Set up Actual Computer (actual.inc) inference in Hermes.

Run 150+ AI apps (image, video, LLM) via inference.sh CLI.

LLM-обогащение готового architecture.json — память агентов, семантические dataflow-связи, назначение модулей; всё бездоказательное merge сам пометит inferred…

Find and rank datasets for a replication or ML task by requirements, license, and availability.

Applies Tom M.

Use when comparing basic, enhanced, GraphRAG, or agentic RAG designs, evaluating domain-specific RAG quality, tuning retrieval components, or deciding whether…

Use when building Hugging Face smolagents, code-executing agents, Python-action agents, Hub-shared tools, or lightweight agent prototypes that need sandboxing…

Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart".

Write, maintain, and migrate CHANGELOG.md files using Keep a Changelog format and best practices.

OWASP LLM Top 10 (2025) audit checklist for AI applications, agent tools, RAG pipelines, and prompt construction.

Use when re-embedding a live corpus after a model, dimension, preprocessing or prefix change: shadow columns, blue-green backfill, cutover.

Design, review, migrate, measure, or debug PostgreSQL vector retrieval with pgvector.

Use when the user directly requests efficiency-aware multi-angle unit tests or a test report after code changes.