Workshop: Shipping LLM systems that survive production
camerongreen95 · reddit · 2026-09-01
Bruno Gonçalves, PhD, is hosting a hands-on masterclass on Sep 12 for engineers building with the OpenAI API, focusing on engineering discipline over "shipping on vibes."
Key Topics:
- Prompt Management: Versioned prompts with regression tests to prevent silent degradation.
- Evaluation Harness: Combining deterministic checks with LLM-as-judge.
- Model Comparison: Bootstrap confidence intervals and paired significance testing.
- RAG: Evaluation with retrieval metrics (recall@k, MRR).
- Robust Agents: Agents with guardrails and fallbacks that fail gracefully.
- Observability: Full tracing, cost/latency monitoring, and CI regression suites.
Related event: Sept 12 Masterclass: Building Production-Ready LLM Systems(2 posts)→
More from coding & agent
- Grok Bot vs. Traditional Coding Tools: Architecture Differences — altryne · 2026-09-01
- Switch from tmux to Herdr for server agent management — HarveenChadha · 2026-09-01
- Shopify shares 'Model Optimization Flywheel' for self-improving LLMs in production — MParakhin · 2026-09-01
- Structured Escalation Cuts Agent Reward Hacking to 5.3% — omarsar0 · 2026-09-01
- Managing remote agents: Control Mac-based agents from your PC — draginol · 2026-09-01
- SWE-bench Multimodal variant focuses on front-end development with vision tasks — OfirPress · 2026-09-01