Design and improve large language model systems from architecture to production deployment.
npx skills add https://github.com/pytholic/claude-skills --skill llm-dev
Please help me install this Agent Skill. Skill: llm-dev Source: https://github.com/pytholic/claude-skills/tree/main/llm-dev Command: npx skills add https://github.com/pytholic/claude-skills --skill llm-dev
This Skill helps you design, evaluate, and ship large language model features without guesswork, covering the full path from model selection to production readiness.
Ask for an end-to-end plan for an LLM feature, including architecture, evaluation strategy, implementation boundaries, and production risks.
High-intent search queries and answers about installing and using this skill.
Build a production-ready LLM system by defining modular architecture, eval-driven development workflows, structured outputs, and safety-aware practices to control latency, cost, and quality. Generate an end-to-end plan covering implementation boundaries and production risks.
Evaluate and improve retrieval-augmented generation by defining eval-driven development plans that measure quality, track metrics, and establish clear failure boundaries for agentic workflows. Apply structured tool workflows and MCP integrations for robust retrieval.
Use fine-tuning methods like LoRA or QLoRA when adapting transformer architectures for specific tasks through preference optimization. Apply pretraining, DPO, and RLHF alignment techniques based on tradeoffs between mixture-of-experts configurations and training objectives.
MCP integrations and agentic patterns support structured tool workflows with clear failure boundaries for large language model systems. Build retrieval-augmented systems with modular engineering to handle tool use, structured outputs, and complex agent interactions safely.
Control latency and cost in large language model deployment through eval-driven development, modular engineering, and safety-aware production practices. Analyze transformer internals, attention variants, and mixture-of-experts tradeoffs to optimize system performance and observability.
pmcdowallAlign large language models via reinforcement learning and preference-based fine-tuning workflows.
daemon-blockint-techGuide end-to-end development of LLM-powered systems with evaluation and observability.
zhangzhang-111-iDesigns RAG pipelines with chunking, embedding, retrieval, and evaluation strategies.
Simon-Copilot-StudioDeploy Python-native ML workloads on serverless GPUs with REST endpoints.
