Deep Learning & Computer Vision
Is My Passion
I'm a Deep Learning Engineer specializing in Computer Vision,
Multi-Agent AI Systems, and MLOps. Currently at Rakuten Mobile,
I've engineered production AI solutions that delivered 40% operational efficiency gains
and 85% reduction in field re-visits through intelligent automation.
My work includes pioneering real-time computer vision systems for telecom deployments,
building agentic AI chatbots using LangGraph and Google ADK, incorporating MCP and A2A protocols for L1 engineering operations,
and publishing research at CVPR 2024 (ToonerGAN). I hold an M.Tech in AI from IIT Jodhpur
with expertise in PyTorch, TensorFlow, Docker, and cloud deployments (GCP, Azure, AWS).
I'm passionate about transforming complex problems into elegant, scalable AI solutions
that drive real-world impact.
My
Projects
 (1).png){kind=logo}
01
AI Agent
AI Roaming Engineer(AIRE) Agentic Flow
Multi-Agent System with Google ADK & MCP
Enterprise-grade Microsoft Teams chatbot automating L1 roaming operations using LangGraph supervisor architecture, Google ADK (Agentic Development Kit), and MCP (Model Context Protocol) with A2A (Agent-to-Agent) communication. Automates test creation, VIP tracking, ticketing, and KPI reporting, achieving 40% operational efficiency improvement.
02
Computer Vision
Cell Fault Detection
Real-time Computer Vision for Telecom
YOLOv11-based object detection system identifying installation errors during cell site deployments. Achieved 83% recall on client data and reduced re-visits by 85%. Containerized with Docker for scalable cloud inference.
03
Research
ToonerGAN
CVPR 2024 Publication
GAN framework generating high-res cartoon avatars to obfuscate identity from automated facial indexing. Trained on 23,000 images (ToonSet), balancing privacy protection with human recognizability. Published at CVPR 2024.
04
Robotics
Autonomous Driving Vehicle
ML-Powered Navigation
Fully autonomous vehicle prototype using Haar-Cascade classifier on Raspberry Pi. Follows road curves and stops at traffic signs with image-feedback logic controlling Arduino for steering (8-bit intensity control via motor driver).
05
GAN
Satellite to Map Generation
Pix2Pix GAN from Scratch
Built pix2pix GAN achieving generator and discriminator loss of 5. Outperformed existing implementations on SSIM score for majority of test samples, converting satellite imagery to detailed maps.
06
AutoML
Neural Architecture Search
Genetic Algorithm Optimization
Applied genetic algorithm for neural architecture search on Fashion MNIST. Discovered optimal CNN architecture with just 11,000 parameters achieving 85% test accuracy through evolutionary genome-based optimization.
My Work
Experience
Data Science Engineer
Rakuten Mobile Inc., Tokyo, Japan
2024-
• Drove 40% efficiency improvement through AI-driven automation of operational workflows
• Engineered multi-agent AI chatbot using using LangGraph and Google ADK, incorporating MCP and A2A protocols for L1 engineering operations
• Pioneered real-time Computer Vision system with YOLOv11, reducing re-visits by 85%
• Technologies: Python, PyTorch, LangChain, LangGraph, Docker, MySQL, YOLO
Graduate Researcher (MTech)
Image Analysis and Biometrics(IAB) Lab, IIT Jodhpur
2022-2023
• Published "ToonerGAN: Reinforcing GANs for Obfuscating Automated Facial Indexing" at CVPR 2024
• Developed novel GAN architecture trained on 23,000 images for privacy-preserving avatars
• Balanced identity obfuscation with human recognizability through combined style + de-identification modules
• Technologies: PyTorch, CUDA, Python, GANs, Computer Vision
Machine Learning Engineer Intern
Skribe
2021 - 2022
• Built web crawlers for 500+ news websites using Python and BeautifulSoup
• Scraped and processed 10,000+ articles with robust pagination handling
• Optimized data collection pipeline for automated content aggregation
• Technologies: Python, BeautifulSoup, Web Scraping
M. Tech in Artificial Intelligence
Indian Institute of Technology (IIT) Jodhpur
2021-2023
CGPA: 7.62/10
• Specialized in Deep Learning, Computer Vision, and Natural Language Processing
• Teaching Assistant for MLOps (CSL7040), Deep Learning (CSL7590), and DLOps (CSP7030)
• Member of Robotics Society, focusing on AI-ML implementation
• Published research at CVPR 2024
B.Tech in Computer Science Engineering
Jharkhand Rai University, Ranchi
2017-2021
CGPA: 8.96/10 (Gold Medal)
• Graduated with Gold Medal for academic excellence
• Qualified GATE 2021 with 98.2 percentile
• Built foundation in algorithms, data structures, and machine learning
What I Offer
Deep Learning Engineering
Real-time object detection, image processing, and custom deep learning CV models for industrial and commercial applications with edge deployment.
Skills & Tools
- Python
- CUDA
- C++
- TensorFlow
- PyTorch
- Albumentations
- Scikit-learn
- Numpy
- Pandas
- YOLO
- R-CNN
- OpenCV
- Matplotlib
- Object Detection
- GANs
- CNNs
- Image Generation
- Docker
- MLOps
- Wandb
- ONNX
- TensorRT
Agentic AI Systems
Building intelligent multi-agent systems using Google ADK, LangGraph, and MCP for autonomous workflow automation and agent-to-agent coordination.
Skills & Tools
- Python
- Google ADK
- MCP
- A2A
- Langchain
- Langgraph
- Microsoft Bot Framework
- Microsoft Teams API
- MySQL
- PostgreSQL
- Rest APIs
- Vector DBs
- RAG
- Agent Architecture
Achievements
& Recognition
CVPR 2024 Publication
Published "ToonerGAN" at Computer Vision and Pattern Recognition conference, one of the top-tier Computer Vision conferences globally.
GATE 98.2 Percentile
Qualified Graduate Aptitude Test in Engineering (GATE) 2021 with 98.2 percentile rank.
Gold Medal - B.Tech
Graduated with Gold Medal from Jharkhand Rai University with 8.96 CGPA in Computer Science & Engineering.
Teaching Assistant - IIT Jodhpur
Mentored students in advanced courses: MLOps, Deep Learning, and DLOps at IIT Jodhpur.
Get In Touch
I'm always open to discussing new projects, opportunities, or collaborations. Whether you have a question or just want to say hi, feel free to reach out!