I'm an AI/ML Researcher and Generative AI Application Developer with hands-on experience spanning Federated Learning, Computer Vision, Deepfake Detection, LLM Security, and Reinforcement Learning.

My research is currently under peer review at Springer's International Journal of Intelligent Transportation Systems Research, and I hold an Indian Patent (filed) for an IoT-enabled computer vision system. On the engineering side, I specialize in fine-tuning and deploying open-source LLMs into production using Python, FastAPI, and MLOps tooling.

Currently exploring: LLM Security & Red Teaming, Agentic AI systems, and scalable MLOps pipelines.

Generative AI & NLP: LangChain · LangGraph · AutoGen · HuggingFace · OpenAI GPT · spaCy · NLTK · RAG Pipelines · Conversational AI

ML / DL & Research: PyTorch · TensorFlow · Scikit-learn · OpenCV · Federated Learning · Reinforcement Learning · Ensemble Methods

MLOps & Cloud: Docker · Kubernetes · CI/CD (GitHub Actions) · AWS · Prometheus · Grafana · ELK Stack

Incident Response Commander – AI Agent RL Environment — Designed a novel RL research environment simulating SRE incident response across a 10-service microservice architecture. Benchmarked a Llama 3.1 8B baseline agent scoring 0.70/0.62/0.56 across Easy/Medium/Hard tasks.

Qubits Learnova – eLearning Platform — Architected a modular Express.js + Prisma + PostgreSQL backend with 13 lesson types, RBAC, Razorpay integration, gamification, and AI quiz generation.

Deepfake Detection System — Researched ensemble deep learning approaches (EfficientNet, InceptionNetV3, ViT) across 5 benchmark datasets totaling 82.6 GB for face-swap deepfake detection.

Ranked 3rd among numerous teams for developing impactful technical solutions under the SSIP initiative.

Recognized as a national finalist in both the March and June 2025 editions.

1. FED-DETR: Privacy-Preserving Intelligent Traffic Enforcement using Federated Learning
   International Journal of Intelligent Transportation Systems Research (Springer) | Under Peer Review
   Developed a decentralized object detection pipeline using Federated Learning and DETR (Detection Transformer) for automated helmet detection in intelligent traffic monitoring.

2. Automated Waste Segregation Dustbin for Educational Institutions
   Indian Patent | Application No. 202521111515 | Under Review
   Designed and prototyped an IoT-enabled smart dustbin with a lightweight computer vision model for real-time automated waste classification on edge devices.

GATE DA 2026 Qualified	AWS Certified Cloud Practitioner

Selected for a competitive 1-month hybrid international immersion program — 6-day academic residency at the University of Kent (London) and 6-day business residency at IESEG School of Management (Paris), covering Global Leadership, Cross-Cultural Collaboration, and International Business Strategy.

I'm always open to research collaborations, interesting conversations, and new challenges — whether it's federated learning, LLM security, MLOps, or anything in between.