Md Roman Nihal

> AI_ENGINEER & RESEARCHER_

View_Work [ GitHub ]
Machine Learning Deep Learning Computer Vision RAG AI Agents LLM Fine-Tuning Machine Learning Deep Learning Computer Vision RAG AI Agents LLM Fine-Tuning

About_Me

I am a Computer Science graduate and AI Developer specializing in building highly scalable, context-aware AI systems.


My expertise bridges the gap between academic research (Deep Learning, OCR, Remote Sensing) and applied engineering (Multi-Agent Systems, RAG pipelines, LLM Fine-Tuning).

Experience

Junior AI Developer
Betopia Group
Feb 2026 – Present
Full Time
Research Assistant
Department of ADS, GUB
Sep 2025 – Jan 2026
Part-time
Machine Learning Intern
Codveda Technologies
Nov 2025 – Dec 2025
Remote
Administrative Intern
Department of CSE, GUB
Apr 2024 – Aug 2025
Student Intern

Publications

Fast and Memory-Efficient 4SQR-Code Decoding using Histogram Analysis
IEEE International Conference on Smart Technologies and Systems for Next Generation Computing (i-COSTE 2025)
M. S. R. Sheam, M. R. Nihal, S. I. Emon, M. R. Alom, K. U. Oyshe, M. A. Rahaman, A. S. Ali
April 2026
[Published]
A 48-Class Handwritten Dataset for the Endangered Chakma Language
Data in Brief
J. Ferdeous, M. A. Kayum, A. Islam, M. R. Nihal, A. Shihavuddin, M. A. Hasan
April 2026
[Published]
Optimizing Lightweight DNNs for Accurate, Low-Energy Object Detection in Remote Sensing
IEEE International Conference on Quantum Photonics, Artificial Intelligence & Networking (QPAIN 2026)
A. I. Nayan, M. A. Y. Nayeem, A. Salam, M. R. Nihal, A. Shihavuddin, M. A. Hasan
Feb 2026
[Accepted]
Self-Supervised Learning for Label-Efficient Wheat Head Detection under Dense and Occluded Field Conditions
IEEE International Conference on Quantum Photonics, Artificial Intelligence & Networking (QPAIN 2026)
M. R. A. Rashid, A. Shihavuddin, A. K. A. Azad, M. A. Hasan, M. A. Niloy, M. R. Nihal, R. Gani, F. R. Sams, R. U. Islam
Feb 2026
[Accepted]

Selected_Works

Mongo-Chat

Stateful LLM-powered chat system integrated with MongoDB for persistent conversations.

Google Generative AI | PyMongo | MongoDB

RAG Lab

Evaluated multiple RAG architectures with vector retrieval and multi-LLM integration via Streamlit.

OpenAI | Groq | Gemini | Vector DB

Agora Scheduler

Autonomous multi-agent system using Google ADK & Gemini 2.5 Flash Lite.

Google ADK | Gemini | Multi-Agent

LLM Fine-Tuning

Parameter-efficient fine-tuning and alignment (LoRA, QLoRA, RLHF, DPO) using Unsloth.

Unsloth | LoRA | RLHF | DPO

Smart Triage AI

High-performance FastAPI endpoints and rigorous Pydantic validation for AI patient triage.

FastAPI | Pydantic | SmythOS
Explore All Repositories

Competitions

Agents Intensive - Capstone Project
Google & Kaggle
[ Kaggle ]
HackTheAI Hackathon (Finalist)
AI Assistant Development using SmythOS
[ GitHub ]
Kaggle Playground Series
Regression, Classification, Time Series Challenges
[ Kaggle ]

Contact_Me

> System Ready. Open for collaborations.

[ Email ] [ LinkedIn ] [ GitHub ]
roman@ai: ~/terminal [X]
Loading AI Assistant module... OK.
Context: Md Roman Nihal's Portfolio.
Type your query and press Enter.
$