researcher = {
"name" : "Ahtasham Ilyas",
"degree" : "MSc AI & Intelligent Systems β University of Bremen",
"research" : ["Machine Learning", "Computer Vision", "3D Perception", "Robotics"],
"project" : "ACRAMbly β Real-time 6D Pose Estimation Pipeline (ROS2 + FoundationPose)",
"award" : "π₯ 1st Place Β· Bremen Big Data Challenge 2025 (Anti-Money Laundering)",
"interests": ["Geometric Deep Learning", "Equivariant Networks", "Generative Models",
"Diffusion Policies", "Multimodal AI", "Interpretable ML"],
"looking_for": "Research collaborations Β· Thesis opportunities Β· Top EU/NA AI labs",
}| Project | What I Built |
|---|---|
| π€ Perception / ACRAMbly | Real-time 6D pose estimation pipeline for robotic assembly β ROS2 Jazzy + Intel RealSense + FoundationPose + GroundedSAM |
| πΈ Anti-Money Laundering | Ensemble ML system for AML fraud detection Β· Won Bremen Big Data Challenge 2025 |
| π¬ VLM-Action Acquisition | Vision-Language Model based action learning with RGB-D preprocessing for robotic tasks |
| 𧬠VQVAE | Vector-Quantized Variational Autoencoder from scratch in PyTorch |
| π MUST Research Portal | Full-stack research management system with PDF reporting |
| π¦ Negativa-ML | GPU utilization analyzer built in Rust |
| π₯ Mammo-Reg | Mammogram image registration for medical imaging |