Dhruv Nigam

ML Engineer & Researcher

I enjoy working at the intersection of math, software, and business. I build systems, not models. Models are the outcomes of well-designed training and feedback systems. Right now, my ambition is to build systems that build and deploy the next generation of AI models.

Recently my team and I launched the first live sports AI influencer which has reached more than a million users. I find myself debugging voice pipelines, optimizing conversational latency, and toying with communication protocols (TCP vs. UDP), model inference engines, and reinforcement learning to make models feel more human.

Previously, I trained my own 200M foundation model to predict user behavior trajectories using a transformer backbone with slight modifications to the standard LLM architecture. Increased DAU by 3% overnight. Validated my hypothesis that scaling laws emerge across different domains, not just text.

Even earlier, I worked as a quant at an investment bank and had billions (not mine) ride on my trading algorithms. There was no room for error, no patience for latency, and real money on the line. That experience taught me the value of rigor in engineering and modeling.

Interested in LLM post-training, inference, and agentic systems.