L&T Technology services

• Worked for a US-based client on Major aspects of a production machine learning system like data training, building models and deploying models, building API services for exposing these models, maintaining them in production, and more. • Spearheaded end-to-end pipeline development for intelligent vision-based solutions on large construction sites, involving multi-camera streams and real-time object detection (Video Analytics) reporting on Power-BI using Azure services. • Proficiently utilized data science toolkits including Jupyter Lab/Notebooks, pandas, and bash scripting, alongside substantial expertise in Linux environments. • Applied practical solutions in diverse domains such as structured information extraction, object detection, Image Classification , Image segmentation, and pose estimation. • Demonstrated adeptness with image-processing libraries such as Python, Scipy, Spacy, Pandas, Numpy OpenCV, PIL, and OCR engines, enhancing image analysis accuracy. • Expertly navigated computer vision and deep learning frameworks and libraries, including PyTorch, TensorFlow, Keras, Scikit-Learn, and ONNX, facilitating innovative model development. • Leveraged cutting-edge algorithms such as YOLO and Faster R-CNN, elevating precision in object detection and scene understanding. • Collaborated within L&T Technology Services GenAI team, actively contributing to a pioneering Large-scale Machine Learning (LLM) application, working alongside Business Vertical Heads. • Played a pivotal role in designing and developing Image Enrichment microservices for client which was consolidated solution of image classification, ROI Cropping, Barcode detection, OCR , Word processing etc., Containerizing (Docker) the solution and making it accessible using REST API.

L&T Technology services

• Collaborated with a prominent global microcontroller leader, specializing in deploying machine learning models onto their SoC, with a focus on microcontroller-based implementation (TinyML). • Executed successful deployment of TFlite micro models onto microcontrollers, including esp32cam, employing advanced pruning techniques for precise object detection. Conducted comprehensive AI model benchmarking. • Championed image processing techniques in the manufacturing domain, harnessing Basler cameras for expedited image processing. This resulted in enhanced image analysis efficiency, contributing to streamlined operations.

Cogent Controls

• Conducted electronic circuit embedded testing and Python simulations, ensuring optimal performance and reliability. • Contributed to the PCB Component Fault Detection team, employing computer vision techniques for efficient and accurate fault detection in electronic components. • Led data collection efforts for diverse PCBs and engaged industry experts to identify critical detection points and faults. • Successfully developed an application capable of comprehensive component detection within PCBs, bolstering quality control measures. • Implemented deployment through Heroku as an API, complemented by a user-friendly front-end dashboard for enhanced accessibility.