I study how neural networks store and forget knowledge. I also build products with those same networks through Factoryze — the gap between understanding something and actually deploying it turns out to be where most of the interesting problems live.
I teach computer science at BRAC University, train Bangladesh’s national AI olympiad team, and occasionally write fiction. Outside of all that: films, anime, games, and good conversations.
Experience
Co-founder · Factoryze
October 2023 – Present
Building AI-powered products end to end — from architecture to deployment. Work includes an STT + generative AI sales platform for a Singapore-based startup, LLM integrations for edtech systems, a document processing pipeline for complex content (images, equations, non-unicode text), and system design for e-commerce, CRM, and edtech using NestJS, FastAPI, PostgreSQL, and MongoDB.
Lecturer · BRAC University
January 2026 – Present
Teaching undergraduate computer science. Courses this semester: Data Structures Lab (CSE220), Computer Graphics Lab (CSE423), Software Engineering (CSE470). If you’re one of my students, feel free to reach out by email for anything course-related.
Academic Trainer & Problem Setter · Bangladesh AI Olympiad
June 2024 – Present
Trained Bangladesh’s national team for the International AI Olympiad. The team won silverware in Saudi Arabia. Topics covered: deep learning fundamentals and why it works, reinforcement learning, training best practices.
Machine Learning Engineer · mPower Social Enterprises Ltd.
December 2023 – July 2025 · Part-time
Fine-tuned computer vision models (YOLO, PyTorch), improving detection accuracy by ~9%. Built RAG, TIR, and prompt engineering pipelines to optimise LLM reasoning via LangChain, SymPy, and Chroma — reducing inference cost through test-time scaling. Developed OCR pipelines for document understanding and researched ASR for accented Bangla.
Trainee Backend Engineer · mPower Social Enterprises Ltd.
November 2022 – August 2023
Built a backend wrapper over a large, mature PostgreSQL database to generate lazy tables and optimised queries via an interactive dashboard.
Education & Research
MSc in Software Engineering · University of Dhaka
Institute of Information Technology · 2024 – 2025 · GPA 4.00 / 4.00
Research on continual learning: models with better stability and plasticity, and how neural networks store and retrieve knowledge and concepts.
Publication — Concept Inconsistency: Using Concept Level Information Learned by Deep Learning Models to Improve Training · 29th International Conference on Knowledge-Based and Intelligent Information Engineering Systems (KES 2025)
BSc in Software Engineering · University of Dhaka
Institute of Information Technology · 2019 – 2023 · GPA 3.94 / 4.00
Notable courses: DSA, distributed systems, system design and software architecture, design patterns, information security. Activities: Competitive programming trainer, AI mentor.
Competitions
Top placements in national ML competitions:
- Top 10 of 240 teams — Robi Datathon 3.0, 2024
- 3rd of 98 teams — DL Enigma 2024, SUST · vehicle detection for autonomous driving
- 2nd of 49 teams — Bhashamool 2024, UIU · Bengali IPA translation
- 2nd of 94 teams — DLSprint 2, Buet CSE Fest 2023 · Bangla document layout analysis
Skills
Languages — Python, TypeScript, C++, Java, Rust
ML / AI — PyTorch, HuggingFace, LangChain, Chroma, YOLO, Whisper
Backend / Web — NestJS, FastAPI, React, Flutter, PostgreSQL, MongoDB
Tools — Git, Docker, Postman, Appium