Building Production LLM Applications

Build, deploy, and scale production‑grade LLM applications: prompts, RAG, agents, fine‑tuning, and serving.

LLM Engineering · Advanced · 10 weeks · 30 lessons · Free

Focus: reliable apps, not demos.

You’ll build:

• Prompting: few‑shot, structured outputs, eval harness.

• RAG: ingestion, chunking, embedding, vector DBs, hybrid search, re‑ranking, and eval.

• Agents: tool use, memory, guardrails, error handling.

• Fine‑tuning: LoRA/QLoRA, data prep, metrics.

• Serving: vLLM, streaming, caching, cost control, observability.

Deliverable: a deployed end‑to‑end AI app you can maintain.

Lessons

  1. Prompting: few-shot, structured outputs, and eval harness (~45 min)
  2. RAG: ingestion, chunking, embedding, and vector DBs (~55 min)
  3. Hybrid search and re-ranking (~40 min)
  4. Agents: tool use, memory, and guardrails (~50 min)
  5. Fine-tuning: LoRA/QLoRA, data prep, and metrics (~50 min)
  6. Serving: vLLM, streaming, caching, and observability (~45 min)