Projects

Financial Document Query AI (RAG System)

AI-powered querying tool for extracting tables and narratives from financial PDFs, using OpenAI embeddings with pgvector-backed PostgreSQL search and custom ranking, resulting in improved analyst efficiency.

Intelligent Timesheet Generation

End-to-end pipeline for automating resource time tracking and reporting, combining Azure OpenAI LLMs, Jira REST API, secure configuration (YAML), and CI/CD for seamless adoption.

Geospatial Traffic Analytics Platform

Real-time dashboard integrating predictive ML models with geospatial APIs (Turf.js, Mapbox GL JS, ArcGIS) for analyzing and forecasting urban traffic patterns; deployed with robust backend (Flask).

Retail AI Agents (Ongoing)

Building bespoke AI agents and multi-agent systems integrating advanced LLMs, designed to automate inventory management, personalized marketing, and customer support within retail settings.

Publications

IEEE CONIT 2024: From Pixels to Prognosis: Exploring from UNet to Segment Anything in Mammogram Image Processing for Tumor Segmentation

Selected for presentation and publication at the IEEE CONIT 2024 conference

Explored state-of-the-art medical image segmentation techniques, including U-Net and SAM, for improved accuracy in tumor detection.

Research Paper Link Certificate Link

Skills

AI/ML & NLP

  • Retrieval-Augmented Generation (RAG)
  • LLMs (OpenAI, Azure)
  • Multi-Agent Systems
  • Text Embeddings
  • Predictive Modeling

Programming

  • Python
  • JavaScript
  • SQL
  • YAML
  • HTML/CSS

Tools & Technologies

  • PostgreSQL (pgvector)
  • AWS (EC2, Lambda, Step Functions, Athena, S3)
  • Azure
  • On-Prem Deployments
  • Git, GitLab CI/CD

Geospatial Analytics

  • Turf.js
  • Mapbox GL JS
  • ArcGIS

Frameworks & APIs

  • LangChain
  • Flask
  • Jira REST API
  • Postman

Evaluation & Soft Skills

  • Precision@K, MMR, NDCG
  • Analytical thinking
  • Problem-solving
  • Clear communication