Shubham Gupta

Implementation of an advanced RAG model to efficiently query and analyze 4000 US government order documents. Developed and deployed a Text-SQL-fact-graph using OpenAI LLMs. Engineered synthetic data sets with realistic distributions for consumer electronics and organic foods markets. Leveraged clustering, RFM analysis, and Apriori algorithms to enhance product recommendations. Spearheaded a proof of concept initiative for demand forecasting and vehicle route optimization. Implemented models using fastapi and docker.

Developed a classification model to distinguish medical prescription (Rx) images from non-Rx images, utilizing neural network models trained via knowledge distillation methodology. Engineered a duplicate image detection system employing the Fastdup library. Orchestrated an active learning loop within Label Studio to facilitate efficient annotation. Completed an audit course on MLOps from Coursera.

Published a paper on "Optimized Routing and Resource Assignment in Quantum Key Distribution (QKD)-Secured Optical Networks" utilizing deep reinforcement learning. Developed a Python simulated environment for QKD-secured optical networks using the RSA-RL library. Implemented research findings on "Fusion Painting", integrating multimodal data fusion techniques with deep learning for 3D object detection. Engineered a multi-language signboard detection system using deep learning.