Reinforcement Learning
MPC vs RL on the Cart-Pole: An Implementation Study
2026-06-05
Hello, I'm
Abebe
Engineer · Researcher · Germany
I'm interested in mechatronics, AI, robotics, reinforcement learning, and model predictive control.
Work
Featured projects
Care AI: Multilingual AI for Dementia Care
AI communication assistant for caregivers in German care homes, helping them connect with migrant residents in their native language using Validation Therapy, personalised memory profiles, and an in-context RL loop that improves with caregiver feedback. Won the AI4Good hackathon.
Production RAG & LLM System
Led a team of 30+ engineers to build and deploy an LLM + RAG pipeline processing unstructured mental health data in 9 weeks. Implemented RLHF feedback loops, structured output schemas, and evaluation pipelines to close test/production gaps.
ENVELOPE: Deep RL for Multipath 5G Traffic Steering
Modelled ATSSS with Multipath QUIC as an RL environment using RLlib, ran experiments on a real-world 5G testbed, and built automation tooling for orchestration and reproducible evaluation.
GPS Sensor Fusion Pipeline
Multi-sensor data fusion pipeline for real-time GPS localisation using Kalman filtering, merging inconsistent signals from multiple sources into a single reliable state estimate.