Workshop on Sep 12: shipping LLM systems that actually survive production
Heat trend
Collecting trend data
The percentage is based on available heat signal, not comment count or independent people.
A hands-on masterclass on September 12, led by Bruno Gonçalves, PhD, founder of Data For Science, will focus on building robust LLM systems for production. The workshop covers essential engineering disciplines for those using the OpenAI API, including versioned prompts with regression tests, a comprehensive evaluation harness combining deterministic checks and LLM-as-judge, and statistical methods like bootstrap confidence intervals. It also addresses evaluated RAG with retrieval metrics, agents with graceful fallbacks, and full production observability, tracing, and monitoring, ensuring systems survive in real-world environments.
There's a hands-on masterclass on Sep 12 for anyone building with the OpenAI API who wants real engineering discipline instead of shipping on vibes.
Covers:
- Versioned prompts with regression tests, so an edit can't silently degrade quality
- A real eval harness combining deterministic checks and LLM-as-judge
- Bootstrap confidence intervals and paired significance testing for model comparisons
- Evaluated RAG with retrieval metrics (recall@k, MRR)
- Agents with guardrails and fallbacks that fail gracefully instead of compounding errors
- Full production observability, tracing, cost/latency monitoring, and a CI regression suite
Led by Bruno Gonçalves, PhD, founder of Data For Science, who trains engineers at Fortune 500 companies on this exact stack.
Link for more details