Poonam Ligade

I architected and owned Jarvislabs.ai’s ML platform, delivering GPU compute to ML engineers and researchers worldwide. I built its React/Next.js frontend, FastAPI/SQLAlchemy APIs, and infrastructure supporting 1,000+ concurrent users, 10K+ GPU hours monthly, and 99.9% uptime.

I created pre-built GPU Docker images with PyTorch, TensorFlow, and Stable Diffusion, turning hours of researcher setup into minutes. I also built LLM and diffusion inference infrastructure, fine-tuning workflows, and Stripe billing that processed $2M+ in transactions.

I improved page load times by roughly 40% through SSR, bundle, and caching improvements, contributing to 25% growth in organic sign-ups. I defined frontend conventions, led a migration, mentored 10+ engineers, and used systematic PR reviews to reduce production incidents.

Following JarvisLabs.ai’s acquisition, I’ve been building agentic AI systems independently, including a voice-based journaling agent using local speech-to-text and TTS, Gemini API reasoning, and multi-step tool-driven workflows. Earlier, I optimized Spark/Scala and HBase data pipelines and reached a Top 0.1% Kaggle Kernel Expert ranking.