Sagar Ravi Bhavsar
Right now I'm building multimodal AI systems at Meta โ teaching LLMs to see, hear, and understand the world beyond just text.
Before this, I spent 7 years making self-driving cars actually see the road. Built perception systems that fused camera and lidar data, detected objects in real-time, and tracked everything moving around the vehicle. Scaled ML training pipelines and eval frameworks that helped autonomous vehicles navigate cities.
Started my career at Juniper Networks building software for enterprise network switches โ back when I thought routers were the future (they weren't, AI was).
WearVQA: A Visual Question Answering Benchmark for Wearables in Egocentric Authentic Real-world Scenarios
with Eun Chang, Zhuangqun Huang, Yiwei Liao, et al. ยท NeurIPS 2025
Sensor Fusion-based Online Map Validation for Autonomous Driving
with Andrei Vatavu, Timo Rehfeld, Gunther Krehl ยท IV 2020
From Particles to Self-Localizing Tracklets: A Multilayer Particle Filter-Based Estimation for Dynamic Grid Maps
with Andrei Vatavu, Melissa Rahm, Suresh Govindachar, et al. ยท IEEE ITSM 2020