Software Trainings · Prompt Engineering track
Prompt engineering treated as software, not folklore.
A six-week track that turns prompting into engineering: system design, structured output with schemas and tools, evaluation harnesses with golden sets and LLM-as-judge, plus the safety, cost, and latency guardrails production calls actually need.
What this track delivers
Why this track is different.
- 01
System-prompt design
Role design, decomposition, few-shot, and the patterns that hold up beyond demos.
- 02
Structured output
JSON schema, function/tool calling, parsing-and-repair pipelines across GPT, Claude, and Gemini.
- 03
Evaluation harness
Golden sets, regression suites, drift detection — the part most teams skip and then regret.
- 04
Production guardrails
Jailbreak defense, PII handling, cost & latency budgets — baked into every call, not bolted on later.
Week-by-week
Syllabus in order.
- Week 1–2LLM foundations
Transformer intuition, API patterns, prompt engineering.
- Week 3–4RAG pipelines
Embeddings, vector stores, chunking, retrieval quality.
- Week 5–6Tool use & agents
Function calling, agent loops, memory, multi-step reasoning.
- Week 7–8Evaluation & safety
Eval harnesses, red-teaming, guardrails, PII handling.
- Week 9–10Production systems
Streaming, caching, cost control, observability, deployment.
- Week 11–12Capstone
Design, build, evaluate, and ship a production AI application.