I'm CTO & co-founder of The Non-Functional Co., where we build AI compliance infrastructure for regulated finance. Most of my time goes into a stubborn problem: making LLM agents reliable and deterministic enough to trust in places where mistakes have real consequences.
Before this I spent 8 years at Bumble in London, going from engineer to Frontend Core Team Lead across platforms serving 50M+ users — with earlier stops at Opera Software in Sweden and AOL in Bangalore. Nineteen years in, the field still finds new ways to make me feel like a beginner. I write about what I learn along the way.
Research: Active Context Compression: Autonomous Memory Management in LLM Agents — an agent that autonomously compresses its own context, inspired by slime-mold exploration. 22% token reduction on SWE-bench Lite with no loss of accuracy.
- Passing taste to AI models — Jul 17, 2026
- The Constraint Spectrum: how the made earns the trust of the born — Jul 15, 2026
- How I'd do engineering for the next year — Jul 14, 2026
- How I am using Vue with Typescript 7 today — Jul 10, 2026
- Building a harness that makes a small LLM reliable — May 20, 2026
More at nikhil-verma.com.