Excellent article, Jessica! You explained the importance of accurate medical image annotation in healthcare AI very clearly and professionally. The way you connected data quality with real-world medical outcomes highlights how critical precision is in AI-driven diagnostics and patient care. As healthcare AI continues to evolve, high-quality annotation will truly become the foundation for reliable and ethical AI systems. Looking forward to reading more of your insights on medical imaging and AI innovation!

I began learning C++ at university, but I never quite got comfortable with it before losing interest.

You have an impressive mix of product management, AI, and technical expertise with strong real-world impact. Your experience in B2C product growth and user-focused innovation will definitely stand out to recruiters.

Solid step-by-step. One operational note from running Tempo at scale: cardinality on span attributes is the silent killer. We had to pull high-cardinality fields like raw prompt text out of attribute labels and into a separate ClickHouse column with sampling. Query latency dropped from 8s to 200ms. Worth thinking about retention policy early.

This resonates. We caught a cross-tenant prompt injection in our document extraction pipeline where a customer's PDF contained text instructing the agent to summarize other data. The agent didn't fall for it, but we added explicit input boundary instructions in our system prompt and run prompt-injection regression tests in CI now. The data exfiltration risk is real even when the LLM doesn't act on the injection.

Treating prompts like code is the right framing. We added a CI hook with Promptfoo that blocks any merge where regression-test scores drop more than 5%. The hardest part wasn't writing the eval set, it was getting the team to maintain it as prompts evolved. Curious if your catalog covers the silent-degradation case where prompts pass eval but drift in real-world distribution.