Jun 2025 – Jan 2026 · 7 mos · Bengaluru, Karnataka, India · On-site

• Designed and deployed demand forecast pipeline for new product launches and predecessor cannibalization, improving product coverage, and increasing forecast accuracy from 86% to 94%.
• Optimized embedding generation services deployed on the CPU enabling high-throughput and low-latency service.

Jul 2023 – Jun 2025 · 1 yr 11 mos · Bengaluru, Karnataka, India · On-site

• Optimized Multi-GPU Training: Spearheaded multi-GPU training using DeepSpeed for large-scale product embeddings, enhancing search relevance on AJIO and JioMart.
• Built and Maintained Spark-based ETL Pipelines: Designed and optimized Spark-based ETL pipelines, ensuring efficient data processing and up-to-date model training.
• High-Throughput Model Deployment: Deployed low-latency model inference services, enabling real-time recommendations and search functionalities.
• Time-Series Forecasting for Retail Analytics: Experimented with state-of-the-art time-series forecasting models for demand prediction and retail analytics.
• LLM-Powered Analytics Chatbot: Developed an agent-based LLM chatbot for data-driven analytics and reporting.
• Regulatory Compliance with AI: Fine-tuned regulatory violation detection models, utilizing pseudo-synthetic data to enhance compliance and accuracy.

Jan 2023 – May 2023 · 4 mos

• Working as a TA for the course of Software Engineering.

Mar 2022 – Jun 2022 · 3 mos

• Worked as TA for the Introduction to Software Systems course, which includes git, basics of creating software systems, frontend development and database management.

May 2021 – May 2024 · 3 yrs

• Domains of research and experience in:
• Textual similarity in monolingual and cross-lingual setting
• Fake news detection and biases in detection models
• Data processing for mass template text generation

May 2022 – Jul 2022 · 2 mos · Hyderabad, Telangana, India

• Worked on improving the object detection and tracking package for their PAL-USB product.
• Fixing duplicate bounding boxes for tiled input in a YOLO object detection model.
• Creating dynamic 3-Dimensional bounding boxes for detection to replace the the 2-Dimensional
• bounding boxes.
• Tested various YOLO models and integrated selected ones into the working pipeline to improve
• tracking FPS.
• Moved post-processing pipeline from CPU to GPU.

Dec 2021 – May 2022 · 5 mos · Hyderabad, Telangana, India

• Worked on synthetic data generation and data augmentation techniques with Dr. Volkmar Scharf-katz, Dr. Lalitmohan Sanagavarapu and Dr. Ponnurangam Kumaraguru.

Jan 2021 – Apr 2021 · 3 mos · Hyderabad, Telangana, India

• Contributed in creating a full stack web application which allowed the user to create game custom game specifications and provide store coordinates using Google Cloud’s Maps Platform.
• Contributed in creating an Augmented Reality(AR) mobile application made using Unity.
• Followed the Agile methodology throughout the development process.