Rohan Bose

Profile Summary

GenAI and Core Java Developer with hands-on experience in Machine Learning, Deep Learning, and Computer Vision. Skilled in developing and deploying AI-driven solutions including LLM-powered chatbots, video anomaly detection, and image classification systems. Proficient in Java, Python, and SQL with strong problem-solving ability and experience delivering scalable, real-world applications.

Technical Skills

Languages: Java, Python, SQL
GenAI: LLMs, Prompt Engineering, NLP
Deep Learning: CNN, RNN, LSTM
Frameworks: Flask, TensorFlow, Keras
Web: HTML, CSS, JavaScript
Tools: Git, GitHub, Colab, VS Code

Projects

Amar AI – Neural Machine Translation Platform

Designed and deployed a Neural Machine Translation (NMT) system for English-to-Bengali translation using the NLLB-200 foundation model enhanced through LoRA fine-tuning. Developed a scalable Flask backend with RESTful APIs, enabling real-time translation requests. Leveraged PyTorch, Hugging Face Transformers, and PEFT techniques to achieve efficient inference and reduced training overhead while maintaining high translation accuracy.

Python • Flask • NLLB-200 • LoRA Fine-Tuning • Hugging Face • PyTorch • NLP • Machine Translation • REST API

Psylent – GenAI Powered Mental Health Support Platform

Developed a GenAI-powered mental health chatbot using LLaMA-3 to deliver empathetic, context-aware responses with automated severity detection. Implemented embedding-based recommendations and LLM-driven conversation summarization to assist counselors in personalized decision-making.

Python · Flask · Groq LLM (LLaMA-3) · NLP · Sentence Transformers

Crop Disease Detection Using MobileNetV2

Built an image classification system to detect diseases in tomato, potato, and corn leaves using the PlantVillage dataset. Applied transfer learning and image augmentation techniques and deployed using TensorFlow and Keras.

TensorFlow · Keras · MobileNetV2 · Deep Learning

Fight Detection System (YOLOv8n + 3D-CNN)

Developed a surveillance system combining YOLOv8n and ResNet3D-18 for spatiotemporal activity recognition. Implemented sliding window prediction and temporal smoothing to reduce false positives and integrated OpenCV for video processing.

YOLOv8 · 3D-CNN · OpenCV · Computer Vision

SarGam – E-Learning Musical Instrument Platform

Developed a Flask and MySQL-based platform with secure authentication, email verification, free trial system, and UPI payment integration for online music learning.

Flask · MySQL · Authentication · Payments

Railway Reservation System

Built a Java Swing application with MySQL backend for train ticket booking, cancellation, and record management.

Java · Swing · MySQL

Experience

Artificial Intelligence Intern – Fretbox

Sep 2025 – Dec 2025

Worked on developing and fine-tuning AI models for a mental health chatbot with focus on conversational accuracy and emotional understanding. Supported data preprocessing, analytics, and deployment readiness.

Deep Learning Summer School Trainee – PDPM IIITDM Jabalpur

May 2025 – Jun 2025

Completed an intensive hands-on program covering Deep Learning, Computer Vision, NLP, and Generative AI using Python, TensorFlow, Keras, and PyTorch.

Content Writer – TeamEverest (NGO)

Feb 2024 – Mar 2024

Created impactful social content and strengthened written communication and storytelling skills.