Aditya Jolly

About Aditya Jolly

AI & Data professional with 4.5+ years of experience building intelligent automation solutions for global enterprises. Specialized in Intelligent Document Processing (IDP) with 3+ years leading Hyperscience development teams at clients like Chevron, PepsiCo, and BNSF Railway — improving extraction accuracy from 65% to 98% and cutting manual effort by 60%.At Tech Mahindra, I owned end-to-end AI model development across structured, semi-structured, and unstructured documents at scale. Currently working as an AI Engineer at Glib, deployed at Cygnet Infotech, where I\'m building a 3-phase RAG pipeline for the CGST Act using AWS Bedrock, Cohere Embed, Anthropic Claude, and PostgreSQL with pgvector — and contributing to a full-stack Unified Office Management System using FastAPI, Flask, and React.Currently finishing my M.S. in Information Science (ML & AI) at the University of Arizona (GPA 3.7), backed by a PGP in Data Science & Business Analytics from UT Austin.My stack: Python · AWS Bedrock · LangChain · pgvector · PostgreSQL · FastAPI · React · Power BI · Scikit-learn · OCR · Deep LearningOpen to Data Analyst, Data Engineer, and AI/ML Engineer roles where I can build things that create measurable business impact.

Experience

1. Cygnet Infotech

   Glib

   Nov 2025 — Present · Ahmedabad, IN

   CGST Act RAG Agent — Built an end-to-end RAG pipeline for the 413-page CGST Act, structured across 3 phases to progressively improve retrieval accuracy. Used Cohere Embed v3 (via AWS Bedrock) to generate 1024-dimensional embeddings, stored and indexed in PostgreSQL with pgvector for semantic search. Integrated Anthropic Claude (via AWS Bedrock) for answer generation with cited sources. Delivered via FastAPI + Streamlit, containerized with Docker. Achieved ~85–90% accuracy.2. Unified Office Management System — Developed backend APIs for the IT Hardware Management module using Flask + PostgreSQL + SQLAlchemy, with JWT auth and role-based access control. Extended work to the frontend, building and integrating React (Vite) components for Parking, IT Hardware, and authentication flows against a FastAPI backend with Alembic migrations and Docker.Python - AWS Bedrock - LangChain - Cohere - Anthropic Claude - PostgreSQL - pgvector - FastAPI - Flask - React - Docker - JWT - Git

Education

• University of Arizona

  Masters In Information Science, Information Technology

• The University of Texas at Austin

  PGP,Data Science and Business Analytics

• IEC UNIVERSITY

  Bachelor of Commerce - BCom, Business/Commerce, General

  2016 — 2019