AI Engineer · Delhi, India
Building retrieval systems, computer vision pipelines, and production machine learning software.
I work on applied AI problems from architecture through deployment—designing graph- and tree-structured RAG engines over complex document corpora, multimodal vision and OCR workflows, and containerized cloud inference pipelines. Alongside independent research, I take on select AI engineering engagements for startups and technical teams.
Selected Work
Representative architectures with 3–5s ambient system previews.
An open-architecture retrieval framework designed for multi-hop technical and enterprise queries where flat vector similarity falls short. Combines isolated Qdrant vector namespaces, RAPTOR hierarchical tree clustering, and Neo4j Cypher knowledge graphs with cross-encoder re-ranking. Deployed with an open workspace and pre-indexed domain corpora so teams can inspect its retrieval behavior firsthand.
Core Areas
Retrieval & Language Models
Multi-stage RAG, RAPTOR hierarchical summarization, and Neo4j Cypher knowledge graphs for complex technical, clinical, and enterprise documents, paired with cross-encoder re-ranking and structured evaluation.
Qdrant · Neo4j · LangChain · Cohere · Ollama
Computer Vision & Multimodal AI
Object detection, surface anomaly segmentation, monocular depth estimation, and OCR pipelines designed for noisy real-world field conditions and assistive devices.
PyTorch · YOLOv8 · U-Net · PaddleOCR · OpenCV
MLOps & Application Engineering
End-to-end delivery from Python training workflows to containerized FastAPI/Flask services on Docker and Google Kubernetes Engine, backed by experiment tracking, drift monitoring, and Next.js interfaces.
FastAPI · Docker · GCP (GKE) · MLflow · Evidently · Next.js
Experience
AI Code & RLHF Evaluator (Independent Contractor)
DataAnnotation- Evaluate frontier language model outputs across Python software engineering, data pipelines, and mathematical reasoning, authoring reference implementations and failure analyses.
AI/ML & Full-Stack Intern
Healthcare Technology Innovation Centre (HTIC), IITM Research Park- Implemented RAPTOR and GraphRAG architectures using Neo4j and LLM summarization over medical research literature and clinical device documentation.
- Built the ACT Lab web platform with automated research publication scraping and vector semantic search, benchmarked deep learning models on imbalanced clinical imaging datasets, and contributed to Arduino-to-cloud telemetry for an organ transit monitoring device.
Teaching Assistant & ML Viva Examiner
Indian Institute of Technology Madras- Assisted instruction and automated evaluation workflows for the Tools in Data Science course covering applied LLM workflows, web scraping, and analytics.
- Mentored student project cohorts in Machine Learning Practice and conduct technical viva examinations for machine learning capstone submissions.
Web Development Intern
NEXT IAS- Redesigned and built responsive course and event registration pages across desktop and mobile devices, lifting event registration completions by 15%.
Education
Nanyang Technological University, Singapore
Master of Science in Artificial Intelligence
Indian Institute of Technology Madras
Bachelor of Science (B.S.) in Data Science and Applications
Delhi Skill and Entrepreneurship University
Bachelor of Science (B.Sc.) in Data Analytics
Publications & Recognition
CureCue: A Smart Adherence Assistant and Reminder System
May – Jun 2025Technical report co-authored with Prof. Neelesh S. Upadhye (Department of Mathematics, IIT Madras) documenting an offline-capable medication tracking pipeline combining YOLOv8 (92.3% mAP@0.5), PaddleOCR, and Gemini 2.0 Flash. Recognized as Second Runner-Up at the BharatGen 2025 Hackathon.