Experience

1. Applied Research Scientist, Thomson Reuters Lab, India
   • Period: August 2024 - Current
   • Legal AI Reasoning and Model Enhancement:
     • Pioneered a legal synthetic data generation pipeline for Process Reward Models (PRMs), creating domain-specific training datasets that improved legal reasoning capabilities on LegalBench benchmark tasks.
     • Evaluated different Test-Time Scaling paradigms for legal reasoning and optimizing inference-time computation.
     • Architected an IRAC (Issue-Rule-Application-Conclusion) Knowledge Graph framework using Thomson Reuters’ Westlaw corpus and court case data, generating high-quality preference datasets that improved legal reasoning alignment in fine-tuned LLMs.
   • AI-Powered Legal Document Update System:
     • Built a comprehensive end-to-end LLM Workflow pipeline to update Word Document with XML parsing and intelligent contextual mapping for paragraph identification and edit, achieving 70%+ success rate while preserving complete document formatting integrity.
     • Engineered human-in-the-loop validation interface with reasoning chains and alert-point mapping, resulting in 60% reduction in manual processing time while maintaining legal accuracy through strategic human oversight and audit trails.
   • DocEvolver: Created an MVP for “Cursor for Word” as an Extension for updating and understanding MS Word Files for Lawyer-Editors
   • Search-and-Replace Agentic System:
     • Architected a multi-agent AI system for automated Word editing with comprehensive validation pipeline (schema enforcement, content integrity, audit logging), achieving 98% accuracy, resulting 65% reduction in manual content revision.
     • Constructed error-resolving agents with function calling and multi-turn reasoning to fix XML issues, implementing few-shot learning and self-healing mechanisms, resulting in widespread adoption across teams, processing hundreds of documents monthly with sub-8 second processing time per section.
   • Additional Tools:
     • Truth Social Monitor: Created a monitoring system for Trump’s Truth Social posts with sub-3 second latency, generating automated alerts for Reuters journalists
     • Page Flipper: Revived version of the Page Monitor extension for website tracking, eliminating Visual Ping subscriptions for the team.
     • These tools provided Reuters with a critical competitive advantage over competitors.
2. Research Intern, Microsoft Research India
   • Period: January 2024 - July 2024
   • Programming with Representations (PwR):
     • Led backend development for Microsoft’s PwR Studio platform, focusing on the Natural Language to Domain Specific Language (NL2DSL) translation system using GPT 3.5 & 4, and developed a symbolic translation pipeline that generates finite state machines structured as custom DSL, achieving 85% reduction in hallucinations.
     • Formulated Rubrics and Evaluations with error correction over DSL which resulted in valid DSL generation increasing from 65% to 95%.
   • Jugalbandi-Studio-Engine:
     • Architected Python-based platform that converts DSL into scalable finite-state-machine-based chatbot applications. This framework reduced development time by 80% and increased accessibility for government agencies and NGOs.
     • The platform was featured in Satya Nadella’s keynote talks, and I represented Microsoft Research in the pilot project, enabling 15+ nontech organizations to develop AI-powered conversational bots.
   • Jugalbandi(JB) Manager:
     • Established a chatbot management platform supporting multiple channels (WhatsApp, Telegram, Web) with multilingual text and voice capabilities; integrated Bhashini Speech models with Azure service failover mechanisms, ensuring 70% faster deployment of new chatbots.
   • Open-sourced work on GitHub:
     • PwR-NL2DSL, a tool to convert Natural Language to Domain Specific Language.
     • PwR-Studio - Studio environment for Programming with Representations
     • Jugalbandi Studio - Open-source chatbot framework
     • Jugalbandi Manager - Chatbot management platform
     • Our complete system has been picked up by Bhashini to build Chatbots across government initiatives.
   • Mentors: Sriram Rajamani, B. Ashok, Akash Lal, Sameer Segal
3. Research Intern, AI Institute, University of South Carolina

   • Period: December 2022 - April 2024

   • Master thesis on Knowledge Enabled Multimodal Ingredient Substitution, built knowledge graph incorporating 27K ingredients and 40K substitution pairs. Enabled precise ingredient recommendations using multimodal and constraint-based searches. Developed an LLM-based query module for the ingredient substitution knowledge graph. Also submitted this work at AAAI-25.

   • Resources: GitHub Repository

     Dataset on Kaggle

     Dataset used in UC Irvine + Stanford Health Hackathon 2024

   • Formulated cross-modal Recipe Retrieval and developed cooking action recognition for recipe analysis, achieving 95% Recall score leading to paper Cook-Gen at IEEE SMC’23.
   • Mentors: Revathy Venkataramanan Dr. Amit Sheth
4. Visiting Researcher at Societal Computing, Saarland University (SIC)
   • Period: May 2023 - August 2023
   • Project: Time and Multispectral Domain Satellite Image Super-Resolution
   • Worked on Satellite Image Superresolution, utilizing Temporal and Multispectral Information for super-resolution. Utilized high temporal frequency of low-resolution images for Wildlife Tracking. Enhanced disaster analysis through high-temporal frequency & low-resolution images using GAN & Diffusion models.
   • Technologies: GANs, Diffusion Models, Computer Vision, Remote Sensing, PyTorch
   • Mentors: Ingmar Weber, Ferda Ofli
5. Research Intern, University of Maryland, Baltimore US
   • Period: October 2022 - April 2023
   • Project: Personalized AI Assistant, funded under HealthCareNLP grant.
   • Worked on developing Personalized Response Generation models using reward scaling over BART & T5, paper K-PERM accepted at AAAI Symposium’24 and improved NUBIA score by 10%. The work has been focused on using Knowledge and Persona for loss scaling resulting better response generation.
   • Technologies: NLP, Information Retrieval, Large Language Models, Conversational Models, Question Answering, Generative AI
   • Mentors: Manas Gaur
6. AI Intern, EdgeNeural.ai, Pune, India
   • Period: June 2022 - August 2022
   • Project: Accelerated inference, optimized models through quantization, CPU/GPU customization. Developed training and optimization pipelines for automatic model training and hosting.
   • Technologies: Optical Character Recognition (OCR), Object Detection (YOLO, SSD), TensorRT, GPU Optimization, OpenVINO, Docker, AWS, PyTorch, TensorFlow
   • Collaborators: Sarvesh Devi, Chidhambararajan, Dhanraj
7. Research Intern, Video Analytics Lab, IISc, Bangalore
   • Period: May 2022 - August 2022
   • Project: Implemented StyleGAN-based architectures for disentangled video interpretation in multiple domains, enhancing image and video generation using Generative Adversarial Networks.
   • Technologies: GANs, Recurrent Neural Networks, PyTorch, TensorFlow, Python
   • Mentor: Rishubh Parihar
8. Research Intern, Visual Learning and Intelligence Lab, Indian Institute of Technology, Hyderabad
   • Period: November 2021 - April 2022
   • Project: Researched Medical Image Processing with Prof. Dr. C. Krishna Mohan. Developed a Novel Architecture for improved classification of low-quality images and unbalanced datasets. Published SFFNet for Panoramic Dental X-ray Segmentation at IEEE APSCON 2023.
   • Technologies: Medical Imaging, Healthcare, Deep Learning, TensorFlow, PyTorch
   • Mentor: R Sai Chandra Teja
   • Collaborators: Dhruv Makhwana, Rohit Pawar
9. Computer Vision Engineer, AI Mage (WETHEKOO)
   • Period: March 2021 - April 2021
   • Project: Developed Fashion Tagging Engine using Deep Learning, optimized and deployed computer vision models on edge devices, improving customer satisfaction.
   • Technologies: TensorFlow, Computer Vision, Siamese Neural Networks, Tagging Engine, Segmentation
10. Software Engineer, Rhizicube Technologies
    • Period: June 2021 - September 2021
    • Project: Oversaw Server & REST API development and Database for Consumer Data Platform Using Golang(Gin). Built a real-time streaming data pipeline using Apache Kafka. Wrote LinkedIn Scrapper and Generalized Organization Website Crawler to scrap data using Selenium & Beautiful Soup.
    • Technologies: Back-End Web Development, Relational Databases, Kafka, MySQL, Data Scraping, Go (Programming Language), Database Design, Selenium, Python
    • Collaborators: Udit Sarin, Yash Goyal