About

AI Engineer with production experience building LLM-powered systems, including a customer support chatbot integrating Claude and OpenAI APIs with Retrieval-Augmented Generation (RAG), plus independent projects spanning NLP, computer vision, and classical machine learning. Comfortable owning a project end-to-end — from data pipeline and model/prompt design through FastAPI/Docker deployment on AWS.

Experience2

  1. AI Developer

    Jun 2026 — Present

    Codiastic Soft Pvt. Ltd.

    Designed and implemented a production customer support chatbot using Retrieval-Augmented Generation (RAG) and LLM APIs, improving response relevance through context-aware retrieval. Integrated Claude API and OpenAI API into the chatbot's conversation pipeline to power natural language response generation. Built REST APIs in Node.js to support the chatbot service and its LLM integration.

    • Improved response relevance through context-aware retrieval.
    • Integrated Claude API and OpenAI API into the chatbot's conversation pipeline.
  2. AI Engineer Intern

    Jan 2026 — Apr 2026

    Codiastic Soft Pvt. Ltd. · Internship

    Built an end-to-end finance AI assistant using 1 local LLM (Ollama), FastAPI, and MongoDB, with persistent conversation history supporting multi-turn financial queries.

Projects5

  • Medical Question Answering System (RAG + LLM)

    Architected an end-to-end Generative AI application using Retrieval-Augmented Generation (RAG): built a pipeline for PDF ingestion, semantic search via Sentence Transformer embeddings, and ChromaDB vector storage to ground medical question answering in source documents.

    • Python
    • LangChain
    • FastAPI
    • ChromaDB
    • Sentence Transformers
    • Llama 3
    • Docker
    • AWS EC2
  • Face Emotion Detection — Real-Time CV System

    Designed and trained a CNN on 35,000+ images to classify 7 facial emotions, achieving ~85% validation accuracy through data augmentation, dropout regularization, and learning rate scheduling.

    • Python
    • TensorFlow/Keras
    • OpenCV
    • CNN
    • AWS EC2
  • LLM Fine-Tuning & Serving Pipeline

    Built an end-to-end fine-tuning pipeline with KNN and data selection strategies, achieving ~12% F1-score gain over baseline.

    • Python
    • Scikit-learn
    • Data Selection
    • AWS S3
  • AI Data Analyst Agent

    Built an end-to-end AI data analysis agent accepting CSV, Excel, and JSON files and answering natural-language queries using LangGraph orchestration, ReAct-based tool execution, automated EDA, statistical analysis, and visualization.

    • Python
    • FastAPI
    • LangGraph
    • React
    • Pandas
    • ChromaDB
    • SQLite
    • Docker
  • Agent Forge - Multi-Agent AI Research and Decision System

    Orchestrated a Supervisor workflow coordinating 4 specialist agents (Research, Analyst, Risk, Report) with tool calling, human-in-the-loop approval gates, SQLite checkpointing, prompt-injection guardrails, and Langfuse tracing.

    • Python
    • LangGraph
    • Gemini API
    • ReAct
    • FastAPI
    • Streamlit
    • SQLite
    • Langfuse
    • MCP
    • Docker

Education1

  • B.E.

    2022

    Kongunadu College of Engineering and Technology

    Electronics & Communication Engineering

Certifications1

  • Artificial Intelligence

    DataMites Institute