About

I am a Research Assistant in AI & Data Science at Quantumweave Intelligence, with over 0.2 years of hands-on experience in computer vision, deep learning, and AI engineering. I have worked on projects such as video analytics, object detection, tracking, and real-world data applications, contributing to the development and optimization of AI models. My strongest skills include Python, C, C++, JavaScript, TypeScript, PyTorch, TensorFlow, OpenCV, FastAPI, Django, and REST API development. I am currently looking to transition into a Backend Developer role, leveraging my technical expertise and practical engineering experience to build scalable and impactful systems.

Experience

1. Research Assistant – AI & Data Science

   Quantumweave Intelligence

   Aug 2026 – Present · Thiruvananthapuram, Kerala

   Contributing to the DVAS computer-vision project covering crowd counting, vehicle detection, multi-object tracking, and drone intrusion detection. Researched and evaluated STNNet and CSRNet for crowd counting and density estimation, including dataset annotation, density-map generation, validation, and inference optimization. Developed and worked extensively on the vehicle detection and tracking pipeline using YOLO, including model integration, GPU inference, detection configuration, class handling, and tracking workflows. Worked on RT-DETRv2 as an alternative object-detection architecture, including its integration and evaluation within the existing vehicle-detection pipeline. Integrated ByteTrack and BoT-SORT for multi-object tracking and generated persistent track IDs, annotated videos, and machine-readable tracking outputs. Optimized video inference through resolution and FPS/frame-sampling experiments, evaluating the trade-off between processing speed and detection/counting perform

2. AI/ML Intern

   Synbrains Solutions Pvt Ltd

   Mar 2026 – May 2026 · Karyavattom, Kerala

   12-week progressive internship covering production AI engineering, backend development, federated learning, and privacy-preserving ML across healthcare and financial domains. Contributed to a clinical AI reporting platform updated report layouts, implemented dropdown improvements, completed scoring, image handling, and sector mapping modules. Applied FastAPI backend integration and Git branching workflows for collaborative feature development. Developed an automated medical report extraction pipeline from specification documents with zero prior codebase. Built structured extraction pipelines using Pydantic data contracts and heuristic regex-based parsing to normalise raw PDF/DOCX report content into validated ReportDraft schemas.

Education

1. Kerala University of Digital Sciences, Innovation and Technology

   M.Sc. in Computer Science (Machine Intelligence)

   2024 – 2026

2. College of Applied Science IHRD, Kozhikode

   B.Sc. in Computer Science

   2021 – 2024

Projects

• PETZZO : Comprehensive Pet Care Platform

  • Django
  • HTML
  • CSS
  • JavaScript
  • MySQL

  Built a full-stack platform connecting pet owners with verified hospitals, veterinarians, and pet shops, featuring AI-driven disease prediction, real-time pet tracking, and appointment booking.

• Manglish WhatsApp Chatbot (Llama-3 / Ollama)

  • Node.js
  • Llama-3
  • Ollama
  • Hybrid AI

  Developed an offline Malayalam English chatbot combining rule-based intent analysis, similarity-based response retrieval, and local Llama-3 inference enabling cloud-independent real-time regional language messaging.

• KuTu: AI-Powered Tour Guide Marketplace

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • AI Agents
  • Gemini API

  Built an AI-powered travel marketplace that connects travelers with verified local tour guides through fixed packages and a reverse bidding system, enabling users to create personalized trip requests based on destination, budget, and preferences.

• AI-Powered E-Commerce Storefront & Catalog Generation Agent

  • Python
  • Agentic AI
  • LLMs
  • Vision-Language Models
  • FastAPI

  Developed an Agentic AI platform that converts natural language prompts into fully generated e-commerce storefronts by autonomously creating store branding, category hierarchy, landing pages, and website content.

• FedGraph-Shield : Hierarchical Federated Graph Learning for Financial Fraud Detection

  • Python
  • PyTorch
  • Flower

  Designed a hierarchical federated graph learning framework allowing distributed banking institutions to collaboratively train a fraud detection model without sharing raw transaction data.

• Adaptive Confidence Modeling for SHADER-Based 3D Human Body Shape Estimation

  • Python
  • PyTorch
  • OpenCV
  • SMPL

  Introduced a dynamic, image-driven confidence model replacing the original global-σ formulation in the SHADER monocular 3D shape estimation pipeline for loose-clothing scenarios.

• Dynamic Connectome Based Prediction of Reality Monitoring Failure Under Sleep Deprivation (Graph Attention Networks)

  • Python
  • PyTorch
  • fMRI Preprocessing
  • Graph Attention Networks

  Designed a dynamic connectome framework using resting-state fMRI to predict reality monitoring failure before overt behavioural symptoms appear with applications to psychosis-risk and fatigue research.

Achievements

• 20–25% lower confidence-prediction error on 9,656 dressed-human images

  2026

  Adaptive Confidence Modeling for SHADER-Based 3D Human Body Shape Estimation

• Designed a hierarchical federated graph learning framework for financial fraud detection

  2026

  FedGraph-Shield: Hierarchical Federated Graph Learning for Financial Fraud Detection

• Designed a dynamic connectome framework using resting-state fMRI to predict reality monitoring failure

  2026

  Dynamic Connectome Based Prediction of Reality Monitoring Failure Under Sleep Deprivation