Specializing in computer vision, deepfake detection, production RAG pipelines, and high-throughput data infrastructure. B.Tech IT student at VIT Vellore.
Geo-Spatial Tracking and Location Obfuscation
Experience in Research and Industry
CGPA @ VIT
Spanning AI,Backend,DevOps
Engineered a hybrid Bi-LSTM + GRU sequence classification network for high-accuracy image and frame-by-frame video deepfake detection. Includes automated face cropping via YOLO and a Streamlit dashboard for real-time video stream verification.
Developed a end-to-end child emotion detection system. Curated and manually labeled a custom dataset of 1,000+ images using Pigeon-Annotator, trained a lightweight YOLO facial-emotion model on TensorFlow, and deployed an alert dashboard.
Built a zero-cloud-cost, privacy-first local AI assistant powered by Ollama. Combines Retrieval-Augmented Generation (RAG) over local document stores, multimodal image understanding, and CPU-optimized Stable Diffusion generation.
Designed and deployed a unified monitoring stack replacing legacy Blynk infrastructure. Built a high-throughput telemetry ingestion service merging time-series data with static metadata registries using outer joins, regex parsing, and multi-frame partitioning.
Created an enterprise document ingestion pipeline converting scanned and digital PDFs into structured data. Integrated Tesseract OCR, PyPDF2 parser, a Natural-Language-to-SQL schema converter, and a Streamlit RAG interface for non-technical queries.
NIT Trichy
Aspire Systems
Qube Cinemas
Mozilla Firefox Club, VIT Vellore
Vellore Institute of Technology
Vellore, India
Oracle Corporation
Issued Oct 2025Machine Learning Engineer · VIT Vellore '26
I'm an Information Technology undergraduate at VIT Vellore (CGPA 9.00/10.0) focusing on practical machine learning systems, computer vision, and backend data engineering.
Rather than relying on canned benchmarks, I enjoy building models from the ground up to solve concrete engineering challenges — whether that's manually annotating a 1,000+ image dataset for child emotion detection (AttentioMate), training Bi-LSTM/GRU neural networks for deepfake video detection (DeepDetector), or optimizing local offline RAG pipelines with zero cloud API costs (EasyLLM).
During my internships at Aspire Systems and Qube Cinemas, I shipped production pipelines that reduced infrastructure operational costs by 70% and enabled automated document intelligence. I am seeking roles in Machine Learning, AI Engineering, Data Science, or Software Engineering.
I am actively open to Machine Learning, Data Science & AI Engineering and high-impact Software / Backend Engineering roles.
Whether you are scaling computer vision pipelines, building RAG systems, or need an engineer who bridges neural networks with production backends — let's connect.