I'm a final-year Computer Science student (AI & Edge Computing) at MIT Art, Design & Technology University, Pune, with hands-on experience building end-to-end ML and LLM-based systems — from RAG pipelines and agentic workflows to deployed cloud applications.

• Currently building AI-powered systems using LangChain, LangGraph, and multi-agent LLM architectures
• Experienced in RAG pipelines, Corrective RAG, agentic planning, and GenAI deployment
• Deployed production systems on AWS (EC2, S3, SageMaker, ECR) with Docker and CI/CD
• Strong foundation in ML, Deep Learning, NLP, and Computer Vision
• Research background in GANs for materials science at CAMMP, and demand forecasting
• Published at IGTT 2026 · Runner-Up at Smart India Hackathon 2025
• Based in Pune, India · Open to remote and hybrid roles

AI LLM Developer Trainee — HCL Tech (Remote) Feb 2026 – Apr 2026

Built and deployed a Corrective RAG (CRAG) pipeline using LangGraph, Groq Llama 3.1, and ChromaDB for clinical QA over an 11,637-page medical corpus. Improved answer relevance from 0.822 → 0.839 and Recall@5 from 0% → 10% over a Naive RAG baseline. Integrated LangSmith observability and a Streamlit dashboard with hallucination detection safeguards.

AI Research Intern — CAMMP, MIT-ADT University (Hybrid) Sep 2025 – Nov 2025

Built a Conditional GAN in PyTorch for ferritic steel microstructure generation. Reduced FID from 120 to 45–55 over 200+ epochs via architecture tuning. Implemented CUDA-optimized training with checkpointing, boosting experiment throughput by 30%.

Data Science Intern — The Developers Arena (Remote) Jul 2025 – Dec 2025

Built an end-to-end ML and Deep Learning portfolio (regression, classification, clustering, CNNs, YOLOv8, LSTMs, Transformers) with deployment via FastAPI, ONNX, and Streamlit.

TensorFlow LangGraph React FastAPI AWS Docker

End-to-end demand forecasting platform with an LSTM model (MAE ~9.8, RMSE ~12.4), a rule-based risk engine for inventory decisions, and a LangGraph agentic layer with Gemini-powered bilingual executive summaries. Deployed on AWS SageMaker + EC2.

📄 Presented at IGTT 2026 — 8th International Symposium on Innovative Global Technology Trends

LangChain Gemini Pinecone Flask Docker AWS EC2

Medical QA chatbot processing 6,661 pages into 52,181 searchable chunks using Sentence-Transformer embeddings and Pinecone k-NN retrieval. Supports concurrent multi-turn sessions via multithreading. Deployed on AWS EC2 via GitHub Actions CI/CD.

XGBoost LangChain FastAPI NLP

XGBoost regression pipeline (R² = 0.74) with NLP-based sentiment/intent analysis and a GenAI follow-up generation module using LangChain. Built for real-world CRM integration with human-in-the-loop safeguards.

• Oracle Cloud Infrastructure 2025 — Certified Data Science Professional
• Prompt Engineering — Infosys Springboard
• Complete GenAI with LangChain & HuggingFace — Udemy (Krish Naik)
• Model Context Protocol (Intro + Advanced) — Anthropic
• Machine Learning with Python (Honors) — IBM
• Computer Vision — NPTEL