Tag: #llm-judge
Writing on GPUs, LLMs, MLOps, Kubernetes — and mindset · 1 posts
LLM Evaluation & Observability: Eval Harness, LLM-as-Judge, Tracing, Regression Prevention (2025)
Claiming a new model "seems better" is not science. The 1 failure mode of 2025 LLM products is "not measuring". Eval harness design, LLM-as-judge pitfalls and calibration, 3-layer observability (Trace/Span/Metric), produ
2026-04-15 · 11 min read #llm-eval#observability#llm-judge#distributed-tracing#langsmith