Dehit Trivedi

Machine Learning Engineer @Toyota Research Institute

Los Altos, CA, US

Toyota Research Institute | Aug 2024 – Present | Los Altos, CAAt TRI, I lead the development of diffusion policy models and real-time predictive systems for autonomous driving and in-vehicle AI agents. My work focuses on trajectory forecasting, multimodal sensor fusion, and deploying scalable ML solutions both on the cloud and edge. Autonomous Driving & Diffusion ModelingDeveloped and fine-tuned diffusion-based trajectory prediction models using PyTorch, Hugging Face, and generative AI techniques.Integrated multimodal sensor data (LiDAR, radar, camera) for accurate path prediction in complex driving environments.Built robust training pipelines with distributed GPU training, advanced hyperparameter tuning, and synthetic data generation for edge-case handling.Deployed optimized models on edge devices using Docker, Kubernetes, and CI/CD pipelines for scalable rollout across Toyota\'s fleet.🧠 Vehicle Intelligence Agent (VIA) – In-Car AI AssistantBuilt a voice-controlled assistant to interpret real-time CAN/OBD-II data, diagnose vehicle issues, and offer contextual responses.Architected a secure, scalable backend using AWS Lambda, DynamoDB, and API Gateway.Integrated retrieval-augmented generation (RAG) with Bedrock agents, combining internal Toyota knowledge with live vehicle data.Developed a robust voice-to-text pipeline for noisy, low-latency in-vehicle environments.

Nirma University, Ahmedabad, Gujarat, India

Bachelor of Technology - BTech