I am currently working as the Team Lead (Computer Vision) of the Product Team at Wobot Intelligence. My daily work includes leading the development of POCs and Products for the Delhi Office.
I have worked on end-to-end design, development and deployment of several Computer Vision Pipelines. Currently, I am working on optimization of Deep Learning models and Deployment at Scale for both CPU and GPU-based deployments. I also conduct technical round of interviews for new recruits.
During a Hackathon at Wobot, I developed the base algorithm for a lightweight model for handwash detection and fortunately, we also won the Hackathon. This later turned into one of Wobot's notable products - Handwash.ai The main purpose of this product is to enforce hand-hygiene compliance, specifically during the COVID-19 Pandemic.
Simple problems should have simple solutions. Use Image Processing wherever possible to reduce compute cost.
Train Robust Deep Learning Models with extensive Data Augmentation using Tensorflow and PyTorch.
Optimize Model Training and Inference with Efficient Data Pipelines, Model Pruning and Quantization.
Deploy Optimized Models at Scale with Tensorflow and Flask APIs for both Edge and Cloud.