← BACK TO THE CAMPUS

Computer vision at scale.

The problems, tools, and decisions behind the work.

PROFESSIONAL WORK · JIO

Real-time systems, end to end.

At Jio, I worked on license-plate recognition, people analytics, and access-control systems. This involved detection and recognition models, tracking, reproducible retraining, and deployment.

Scaling a vision pipeline is also a systems problem. NVIDIA DeepStream supported efficient camera-stream processing, while Milvus provided vector search for feature-based lookup.

PyTorchDeepStreamOpenCVMilvus

PROFESSIONAL WORK · SAMSUNG ELECTRO-MECHANICS

Understanding driver attention.

I also developed perception models for driver monitoring, including face detection and identification, facial landmarks, gaze estimation, and drowsiness detection.

RESEARCH INTERNSHIP · IIT BOMBAY

Vision for rice phenotyping.

My earlier work at the Agro Informatics Lab involved differentiating rice genotypes through image features, machine learning, and improved annotation workflows.

View the publication ↗