How to Tell If a Model Is Good Enough
Learn why accuracy is a trap, and how precision and recall reveal what really matters when evaluating machine learning models.
7 min read
§ 00Engineering · Writing · Cooking
Nick
Karnik.
Nearly three decades of building software across search engines, disease models, data platforms, and AI tools. Serious about cooking, cocktails, travel, and building games with my three kids. Still figuring out what comes next and building it anyway.
The Modern Attack Surface: How Computers Get Compromised
An overview of how the attack surface has evolved in modern computing, and the ways computers and accounts get compromised today.
17 min read
A Practical Way to Think About RAG
A grounded mental model for Retrieval-Augmented Generation, with two concrete examples and the tradeoffs that actually matter.
7 min read
02Work
01 —
AI and Developer Tools
Built Gemini Code Assist at Google. Writing about what I learned building it.
02 —
Search and Relevance
Six years on Bing search at Microsoft. A patent. A lot of labeled data.
03 —
Making Things
Currently building new tools. Making games with my three kids. Always something cooking.
04 —
Engineering Leadership
Led teams at Google, Microsoft, Tableau, and T-Mobile. Hired and scaled engineering organizations.