Blog
Short, practical writing on building AI-enabled software that survives production — and on becoming an engineer who can build it.
· 3 min read
Why most GenAI projects stall after the demo, and the six things — structure, provenance, review, evaluation, recovery and cost — that make an LLM feature survive real users.
Applied GenAILLM engineeringProduction
· 3 min read
A practical way to choose between deterministic software, an LLM workflow and a tool-using agent — and how to constrain an agent's authority when you do need one.
Agentic AIArchitectureAI safety
· 2 min read
What separates a finished college project from evidence that you can contribute to an engineering team — and a simple scorecard to review your own work.
CareersSoftware engineeringStudents