@mituljn_14
StudentArtificial Intelligence Intern @ WeIntern | AI/ML Engineer | Computer Vision | Generative AI | RAG Systems | LLMs | Agentic AI | MERN Stack
AI/ML engineer-in-training with a 9.11 CGPA and 8+ shipped projects, specializing in Computer Vision, Agentic AI, and RAG systems. Built an industrial defect-detection agent for Tata Technologies (97.76% inspection accuracy), an on-premise multi-agent enterprise platform (91% retrieval accuracy over 700K+ vectors), and a RAG tutoring system ( 1.3s latency). 1 peer-reviewed publication in IJCDS.
• Architected a fully on-premise, zero-data-egress multi-agent platform (LangGraph, Qdrant, FastAPI, Docker) routing queries across 4 specialist agents (document, code, vision, meeting). • Delivered 91% retrieval accuracy via hybrid dense + BM25 search over a 700K+ vector store (7,000+ documents); sandboxed code execution and department-level access control. • LLM-agnostic– swappable to any open-weight model (LLaMA, Mistral, Gemma) via Ollama with zero code changes
• Deployed an agentic YOLOv8 Visual Inspection Agent (OpenCV, Python) that localizes, classifies, and routes industrial defects in real time– auto re-inspecting, escalating, or passing each detection, replacing manual review entirely. • Achieved 97.76% on-floor inspection accuracy (tracked via mAP, IoU, Recall) despite class imbalance, sensor noise, and variable lighting; built end-to-end from camera feed to structured decisions, fully on-premise.
Tata Technologies
• Deployed an agentic YOLOv8 Visual Inspection Agent (OpenCV, Python) that localizes, classifies, and routes industrial…
WeIntern Pvt. Ltd.
• Contribute to AI/ML projects and real-world technical tasks, researching and implementing AI techniques and domain spe…
Computer Science and Engineering (Artificial Intelligence and Data Science)
