Kilo Code
Scaling an AI coding assistant processing over 1 trillion tokens per week
The Problem
Kilo Code set out to build an open-source AI coding suite as fast as possible, covering everything from IDE extensions and CLI to code reviews, enterprise features, and more. Guided by a philosophy we call Kilo Speed: effortless, joyful flow without dependencies, blockers, or friction. Reaching that level of scale and reliability requires more than good engineering. Data protection becomes critical when handling user code, infrastructure must be rock-solid, and team processes need to keep up with rapid growth. Multiple critical functions needed attention simultaneously.
The Approach
Took on a multi-hat role as Senior Engineer and Engineering Lead, combining hands-on development with cross-functional leadership. Managed engineering workflows, established data protection practices as DPO, drove SOC 2 compliance using Vanta, and built out IT infrastructure. Drove architectural decisions to handle the scale of >1T tokens/week reliably.
Tech Stack
Team / References
Highlighted members I worked closely with
Results
- Scaled platform to >1T tokens/week, reaching #1 on OpenRouter
- Established engineering processes and team workflows
- Implemented data protection compliance as DPO
- Achieved SOC 2 compliance using Vanta
- Built reliable infrastructure for high-throughput AI operations