NANDANPATEL

ISRO SAC

Machine Learning Intern

• —Completed development of internal research website using Java Servlet backend with DAO architecture pattern
• —Developed PCB Structure Detection system using Segment Anything Model, Canny edge detection, and skeletonization to produce clean structural boundary overlays for hardware analysis
• —Researching synthetic PCB defect image generation using Diffusion Models, cTransGAN, and SAGAN — multi-architecture comparative research for data augmentation at scale

CoE BVM

Artificial Intelligence Intern

IITD-FSM Internship, AIA Foundation

• —Built a RAG-based intelligent chatbot delivering domain-specific responses for digital manufacturing queries using LangChain and Llama-3
• —Integrated ChromaDB vector database for context-aware document retrieval with sub-second response latency
• —Developed FastAPI backend with full speech recognition and synthesis for hands-free voice interaction in lab environments

EnergInAI

Machine Learning Intern

• —Engineered a novel Dual-Subnetwork NILM architecture integrating Self-Attention and parallel Dilated Convolutions for precise appliance-level energy disaggregation
• —Packaged the Seq2Point model as a production RESTful API mapping aggregate mains power to individual AC consumption and operational status
• —Architecture enables an estimated 15% energy savings for commercial clients through reliable peak demand management

Dept. of Telecommunications, Govt. of India

5G Innovation Hackathon

Developed a complete biomedical signal orchestration pipeline. EMG and EOG signals extracted from body-deployed sensor nodes (Arduino R1 + ESP32), streamed in real-time to a Raspberry Pi, classified using signal processing algorithms, and used to actuate a robotic arm with precise movement control — fully hands-free, intent-driven.

National Robotics Competition

Robofest 4.0

Led full system architecture and embedded integration of KRAKEN, a fully functional underwater ROV. Implemented CV-based barcode detection using OpenCV with real-time web interface communication between Flask server, Raspberry Pi controller, and ESP32 sensor nodes via I2C and UART protocols.

Team GCET Robocon

ABU Robocon 2024

Contributed computer vision and ML pipeline development for the national-level ABU Robocon competition. Built real-time object detection and tracking systems enabling precise robot navigation and task execution under competition conditions.

Deep Learning / Edge AI

Skin Cancer Classification

Feb 2025 — Apr 2025

ResNet50-based deep learning model for multi-class skin lesion classification using transfer learning. Optimized with image augmentation, early stopping, and systematic hyperparameter tuning. Deployed as TensorFlow Lite model for low-latency inference on resource-constrained edge devices. Results published in peer-reviewed research.

Computer Vision / GAN

Virtual Try-On System

AI-powered virtual clothing try-on system combining MODNet for precise background matting with GAN-based garment warping and deformation. Real-time rendering achieved using MediaPipe for accurate body landmark detection with dual-framework inference pipeline delivering sub-100ms response.

Client Website

Radhe Dental Clinic

Multispeciality & Implant Center

Feb 2026

Full production website for a multispeciality dental and implant center. Responsive layout, service presentation, appointment flow, and SEO-optimized structure. Delivered on deadline and currently in active use.

Full Stack MVP

Automate Pro

Enterprise Document Automation

January 2026 - Ongoing

MERN stack MVP for enterprise document processing. Two processing bots: Item Document Processing (manual/SFTP/FTP upload, queue management, Gemini OCR, part number extraction, LN validation, Infor IDM upload) and AP Invoice Bot (cost-based and order-based processing, ION API validation, invoice registration). Includes RAG bot for process Q&A. N-superuser auth system with 2-hour expiring access tokens.

Android Development

Starlight Android Store

E-Commerce Application

Apr 2024

Full-featured Android e-commerce application built in Java. Complete product listing, cart management, checkout flow, and Firebase-backed authentication with cloud storage for product assets and user data.

Published Research

Peer-Reviewed Journal

Research paper on multi-class skin cancer classification using ResNet50 transfer learning accepted and published.

Robofest 4.0 Finalist

2.5 Lakh INR Grant

Led KRAKEN underwater ROV team to national finals securing competitive research and development grant funding.

5G Hackathon Winner

1 Lakh INR Seed Funding

First place and seed funding for biomedical signal orchestration pipeline at DoT, Government of India hackathon.

SAP Advanced Certified

April 2025

Advanced certification covering Machine Learning, IoT, Deep Learning, Computer Vision, and ABAP development.