AR

Hello, I'm

Abdullah Al Rafi

Software Engineer specializing in Machine Learning, Data Science and Full-Stack Development. Currently pursuing MSc in Natural Language Processing.

Trier, Germany

Scroll

About

A bit about me

I'm a software engineer with a passion for building data-driven products and solving complex problems with machine learning. With professional experience at bKash — the largest mobile financial service in Bangladesh — I've worked on production ML systems that serve millions of users.

Currently, I'm pursuing my MSc in Natural Language Processing at Universität Trier, Germany, deepening my knowledge in computational linguistics, deep learning and language technologies.

I enjoy the full spectrum of development — from training models and building APIs to crafting intuitive frontends. Outside of code, I'm curious about cyber security, finance and open-source.

Education

MSc in Natural Language Processing

Universität Trier, Germany

2025 – Present

BSc in Computer Science & Engineering

North South University, Dhaka

2017 – 2022

Interests

NLP & LLMsDeep LearningOpen SourceWeb DevelopmentFinance

Experience

Where I've worked

Senior ML Engineer, Data Science

bKash Limited
May, 2025 – Oct, 2025
  • Port and extend Customer Churn, Business Anomaly and Customer Lifetime Value projects on Apache Airflow.
  • Improved performance of the Birth Certificate OCR solution by 8% with new types of document support which increased weekly successful customer on-boarding up to 3x.
  • Collaborated on business intelligence bot development using Google ADK, vLLM and MCP.

ML Engineer, Data Science

bKash Limited
Apr, 2023 – Apr, 2025
  • Developed an end-to-end multilingual Bangladeshi Birth Certificate OCR solution, first of its kind.
  • Served the solution with NVIDIA Triton and ONNXRT to scale for 100k+ hits per week.
  • R&D on LLM to help the customer support team with primary focus to use LLM without doing any fine-tuning i.e RAG, GraphRAG and mixture of experts.

Junior Data Scientist, Data Science

MedAi Limited
Apr, 2021 – Mar, 2023
  • Created a gold standard dataset in medical domain.
  • Worked with NVIDIA NeMo to fine tune with Bengali ASR dataset.
  • Deployed Django and FastAPI backends on the cloud with fully automated CI/CD pipelines using GItHub Actions and Jenkins.

Skills

Technologies I work with

Languages

Python
JavaScript
TypeScript

Frameworks

Django
FastAPI
React
Next.js
Node.js

ML / AI

Hugging Face
PyTorch
Nvidia Triton
Apache Airflow

Databases

PostgreSQL
MongoDB
Redis

DevOps & Cloud

Docker
AWS
Grafana
Git

Frontend

React
Tailwind CSS
Next.js

Projects

Things I've built

bKash ML Pipeline

Production machine learning services serving millions of customers at Bangladesh's largest MFS.

PythonNvidia TritonREST API

NLP Research

Natural language processing research projects exploring transformer architectures and multilingual text understanding.

Hugging FacePyTorchNLP

Blog Website

This website — a blog built with Next.js 16, Tailwind CSS and TypeScript.

Next.jsReactTailwind CSS

Deep Learning Experiments

Collection of deep learning experiments and implementations covering transformers and generative models.

PythonHugging FaceDeep Learning

Contact

Get in touch

I'm always open to interesting conversations, collaboration opportunities, or just a friendly hello. Feel free to reach out!