At Griffyn Robotech Pvt Ltd, I design and deploy computer vision pipelines for industrial defect detection and quality inspection, reducing manual QC effort by ~90% across multiple client deployments. My YOLO-based detection and segmentation models achieved >95% mAP on client production datasets.
I optimized inference latency by 60% using TensorRT/ONNX conversion and batch processing, and integrated vision modules with PLC automation via PyADS. I also assess factory-floor constraints and translate operator problems into ML formulations.
For the Screw Hole & Thread AOI Inspection System project with Tata Electronics Pvt Ltd, I built a machine vision application that measured production components within 5 microns and trained a segmentation model for thread defect detection. The project included on-site validation during FAT and SAT phases.
Earlier, at Bryckel AI, I developed a lease abstraction solution using RAG and Azure OpenAI, with a LangChain chat interface. At Forvia Hella India Automotive Pvt. Ltd, I developed models for battery state-of-health prediction and vehicle access and identification.
