Principal Software Engineer · Rust + TypeScript · LLM infrastructure
20+ years building production software. Former VP of Engineering at Gojek. Based in Bengaluru, working remotely with teams worldwide.
Rust · TypeScript · React · llama.cpp · Tauri · OAuth 2.1 / Keycloak · Cloudflare Workers · PostgreSQL
What I'm building: BodhiApp
An open-source local AI gateway. It runs open-weight LLMs on your own machine behind OpenAI-compatible APIs, so your apps and your data stay private.
- Rust core that embeds llama.cpp for local inference and routes across multiple AI providers with fallback
- OAuth 2.1 / Keycloak-secured APIs, so third-party apps can use a person's local models only with their consent
- TypeScript/React UI, packaged as a desktop app with Tauri
- bodhi-js SDK and create-bodhi-js scaffolder, so web apps and Chrome extensions can talk to a local Bodhi
- Built over 2.5 years: 2,300+ commits, 180+ releases, 2,000+ downloads
- utilityai/llama-cpp-rs: Rust bindings for llama.cpp. Merged PRs pass build metadata through to downstream crates, fix the unicode.cpp build, and fix the macOS tests.
- gojek/proctor and huggingface/diffusers: smaller contributions
I've been a super user of AI coding assistants since GitHub Copilot first launched, and I've adopted each new generation of these tools as it arrived. I use AI to move fast, but I don't vibe-code. I own the architecture, the tests and the review. Every change starts from clear acceptance criteria, is driven by tests, and lands as a small, reviewed PR. The result is AI speed with code that is clean, well-architected and built to last.
- VP of Engineering, Gojek (2016–2021)
- Founder, MemexAI Technology (2021–present): building BodhiApp and consulting
- Won the Cohere hackathon (2022), judged by Aidan Gomez
- Created a Udemy course on Stable Diffusion (~3,000 students, rated 4.5+)
- Full-time: Senior/Principal IC roles in Rust, TypeScript and LLM/AI infrastructure
- AI-native engineering: helping teams adopt AI coding assistants without trading away quality
- Contracts: architecture, Rust/TypeScript builds, and local LLM inference integration
Connect with me on LinkedIn. Please include a short note about the opportunity with your request.