Rezaul Abedin

I worked on AI projects for manufacturing. I built a pipeline that cleans and structures raw event logs from Volkswagen production lines. I built a transformer model that predicts machine power use from CNC data, and got it running on the oktoflow platform. I also trained reinforcement learning agents (PPO, DQN) for factory scheduling, and tried using BERT to turn shop-floor data into text the model can read.

Data collection, preprocessing, and preparation of industrial production data for machine learning applications.

Provided hardware, software, and network support to end users at Abbott and ALSTOM Transport Deutschland GmbH. Troubleshot technical issues, installed and configured systems, and supported users with day-to-day IT problems.

I managed databases and analyzed sales data to track performance across sales zones.

I managed databases, monitored transactions, and trained staff on banking software across several branches.

Master Thesis : For my thesis I built a simulation of a factory floor with different machines and orders. I wrote a Q-learning agent by hand and trained a DQN agent too, and compared both to a simple rule-based scheduler. Research Project: I worked with a large Bosch production dataset. I looked for the most useful features and tested ways to pick a smaller, representative sample of the data, using XGBoost, Random Forest and SVM.