Rohan Mulay

AI/ML Engineer with 12+ months of experience across 3 startups, specializing in Edge AI, ADAS, and Computer Vision. I'm passionate about building perception systems that make roads safer — taking models from training to production on edge devices in real-world, safety-critical environments. I've contributed to Level 2+ ADAS development across two-wheelers, dashcams, and autonomous tractors, and was part of the core team that built the first working prototype of an autonomous electric tractor from the ground up. Here's what I bring to the table: → Built multi-camera sensor fusion pipelines for forward collision warning and blind spot monitoring, deployed on NVIDIA Jetson Orin NX at sub-30ms latency and 35+ FPS → Developed driver facial recognition (97%+ accuracy) for drowsiness detection and OCR pipelines processing 500+ frames/min for license plate recognition in dashcam ADAS products → Shipped semantic segmentation models with 92%+ mIoU for lane detection on autonomous electric tractors, compressing model size by 60% for embedded deployment → Built LSTM ensemble models for battery SoC/SoH prediction, achieving under 3% RMSE on multivariate time-series data → Full-stack ML toolkit — PyTorch, TensorFlow, TensorRT, ONNX, YOLOv11, OpenCV for modeling; FastAPI, MongoDB, Pinecone, Qdrant for backend and RAG pipelines B.Tech in Computer Science from PES University, Bengaluru (2026). Open to full-time roles in Computer Vision, Edge AI, and Applied ML — always happy to connect!