AI Engineering

Understand how LLMs really work — tokens, transformers, sampling — then master prompting and context engineering to build robust LLM apps.

Build production retrieval-augmented generation — from your first RAG pipeline to hybrid search, reranking, GraphRAG and agentic retrieval, with evaluation.

Build AI agents from scratch — the agent loop, workflow patterns, tools, planning, memory, MCP, multi-agent orchestration — then map them onto frameworks.

The eval-driven discipline used inside frontier labs: error analysis, LLM-as-judge, the improvement flywheel, and production observability.

When fine-tuning beats prompting/RAG — and how: memory math, LoRA/QLoRA, dataset engineering, DPO/GRPO and model merging, with a hands-on run.

Take a prototype to a fast, cheap, safe, well-designed, governed and observable production system — including AI/Agent UX, multimodal, and the frontier.

Three portfolio-grade builds (RAG assistant, MCP multi-tool agent, fine-tuned model) plus an AI-engineering interview and system-design framework.