o First engineer on a ‘zero to one` product identifying high-risk insurance claims. o Built a RAG-based chatbot for product discovery, improving alignment and de-risking the product.o Owned end-to-end development of models for predicting Tail Value-at-Risk.

o Led the end-to-end development of 3 new cold start and automated bidding products, reducing advertiser concentration by 35% across $400M+ in ARR, and increased revenue by $10M. o Proactively managed causes of intra-team conflicts & burnout, resulting in higher revenue and an increase in team satisfaction from 30% to 80%.o Spearheaded raising engineering best practices through mentoring, and led testing improvements that reduced P0 frequency by 90%.

o Developed IMU-based activity recognition models for a wearable fitness prototype. Successfully advocated for pivoting the existing product roadmap to prioritize de-risking the product strategy, accelerating time-to-market by 100%.o Prototyped LLM training, acceleration & deployment, reducing inference time by over 70%.o Identified key risks in data collection across high priority products and proposed solutions that significantly improved product outcomes and development velocity.

o Advised 15 graduate students by designing industrial research projects, setting development goals, coaching, providing performance reviews, and collaborating on academic research. Served as primary advisor for 4 students.o Spearheaded research project for maximizing revenue at a windfarm with battery storage for a Global Fortune 500 client. The solution used an ensemble of forecasting models, stochastic optimization and reinforcement learning, surpassing client’s in-house performance. o Developed MVP of an automated scoring & player analysis tool for table tennis videos using human pose estimation and event detection. Deployed models with Triton Inference Server, TensorRT & Terraform, and built the front-end website.

o Developed state-of-the-art conversational AI, winning runner-up at NeurIPS Alexa chatbot team competition.o Consulted startup executives & ML Teams on AI strategy & implementation, o Led a startup’s 5-person computer vision team through technical challenges to deploy object detection models to the edge.o A Deep Reinforcement Learning Chatbot [NeurIPS Best Demonstration Award – Runner-up, 2017]

o Implemented CNN ensemble that improved accuracy by 90% for tilt prediction in adaptive optics.