My journey began in the picturesque village of Kuppankuzhi, nestled in the vibrant landscapes of Tamil Nadu, India, where I was born and nurtured.

Academic Odyssey

Fueled by an innate curiosity and a passion for innovation, I pursued a Bachelor’s degree in Electronics and Communication from Annamalai University, Chidambaram, laying the cornerstone of my academic odyssey.

Eager to expand my horizons, I transitioned to the United Kingdom to immerse myself in advanced technological realms. At the University of Sussex, Brighton, I honed my expertise through a Master’s degree in Embedded Digital Systems. Further driven by a desire to merge technology with cutting-edge intelligence, I embarked on a second Master’s degree in International Biometrics and Intelligent Vision at Créteil, Paris, refining my proficiency in the art of futuristic solutions

Technical Expertise in Deep Learning, Computer Vision, and Data Science

3D Face Mesh Modeling for Surgical Planning

Currently leading the development of state-of-the-art 3D face mesh models for surgical planning, leveraging deep learning and geometric transformation algorithms. This work integrates 2D face to 3D mesh, pose estimation, mesh optimization, and real-time rendering to deliver precision and utility in clinical applications.

Computer Vision: Object Detection, Tracking, and Recognition

Extensive experience in creating and deploying advanced object detection and tracking models using convolutional neural networks (CNNs), YOLO (You Only Look Once), and Transformer-based architectures. Implemented scalable solutions for real-time recognition tasks with high accuracy, addressing complex challenges in feature extraction, multi-object tracking, and classification.

Sports Analytics Models for Tennis and Cricket

• Tennis Court Homography: Engineered a court homography model to map camera views into a consistent 2D plane using homography matrix estimation, aiding advanced match analysis and visualization.
• Cricket Highlights Generation: Designed a highlight generation model using temporal action detection and frame classification techniques to automate the creation of engaging sports summaries.

Data Science Applications in Forecasting and Segmentation

• Sales Forecasting: Developed robust time-series models for weekly, biweekly, and monthly sales predictions using ARIMA, Prophet, and LSTM networks. These models offered actionable insights for seasonal demand in the shoe and apparel industries.
• Footwear Attribute Segmentation: Built a semantic segmentation model utilizing UNet and Mask R-CNN to classify footwear attributes such as laces, soles, thread stitching, and logos, enhancing product cataloging and customization capabilities.

Real-Time Vehicle Axle Detection System

As a freelance project, designed and deployed a real-time vehicle axle detection and counting system on NVIDIA Jetson Xavier. Leveraging edge computing, TensorRT, and optimized deep learning inference engines, this system currently operates at the USA-Mexico border, enabling precise axle detection for toll evaluation.

Speech Processing and Natural Language Processing (NLP)

• Developed models for speech detection, speech-to-text transcription, and time-aligned speaker diarization using RNNs, Transformer-based architectures (BERT, Wav2Vec), and advanced sequence-to-sequence models.
• Implemented Part-of-Speech (PoS) tagging and Named Entity Recognition (NER) to resolve conversational queries like “Who speaks when?” for real-world applications in transcription services.