About Me
AI Engineer driven by optimizing end-to-end intelligent systems. Focused on building production-grade autonomous workflows that fuse NLP, Large Language Models (LLM), Retrieval-Augmented Generation (RAG) systems, and agentic AI innovation with computational efficiency to solve high-impact problems at scale.
Skills & Technologies
Work Experience
AI/ML Engineer
Built Enterprise SAP OData Gateway using FastAPI and LangGraph, reducing data retrieval time from hours to sub-5 seconds. Developed an Ephemeral Air-Gapped Python Sandbox with Docker for automated SAP data extraction, cutting enterprise reporting latency by 90%. Deployed Next.js 15 conversational interfaces with real-time SSE streaming, reducing LLM token consumption by 85%.
Prompt Engineer & Data Analyst
Engineered Python validation pipelines and automated schema checks, reducing QA time from 4 days to 2 hours. Applied LLM prompt engineering to automate data cleaning and built a geographic code-matching utility for 87% faster reconciliation. Developed Power BI dashboards translating complex data pipelines into actionable insights for stakeholders.
Projects
TechHubAI
Built a multi-agent debate system using FastAPI and Next.js, featuring 4 different persona agents debating tech trade-offs and a judge synthesizing verdicts. Achieved 95% token reduction via research pre-summarization and sliding-window compression, with the system live on Vercel.
Career-Orbit
Developed 6 AI agents using AgentScope and Python that crawl LinkedIn and Wellfound, scoring CV-fit and drafting recruiter emails. Integrated SurrealDB Graph RAG for local-first tracking, with human-in-the-loop form filling via Playwright.
TalentAI
Created an AI recruitment platform using Python and Flask with Gemini Pro & Groq Llama 3.1 for PDF resume parsing and scoring. Implemented weighted ranking for skills, experience, and academics, along with full CRUD HR workflows and applicant tracking.
AI News Analyzer
Developed a multi-model NLP toolkit for fake news, hate speech, and news category classification, deployed as a live Streamlit app. Achieved 89% accuracy for fake news classification and 80% for hate speech detection.
Education
B.Tech
Computer Science