IEEE IV 2025
Decentralized Reinforcement Learning for Multi-Agent Navigation in Unconstrained Environments
Felix Förster, Qadeer Khan, Daniel Cremers
Deep Learning • Multimodality • Scaling Laws
Felix Förster
M.Sc. Computer Science Student | Deep Learning Researcher & Engineer
Deep learning across vision, audio, and language, with a particular interest in learned representations.
News
Jul 2026
I'm happy to join Sophia Koepke and Alexei (Alyosha) Efros at BAIR, UC Berkeley for my research stay this summer.
Oct 2024
Started the M.Sc. Computer Science program at the Technical University of Munich.
Publications
Projects
2026 ML Benchmark for VBF Higgs-Pair Events
Felix Förster*, Lars Schneider*, Johannes Mesner*, Lars Linden, Celine Stauch
Benchmarking 12+ ML methods, including DeepSets, to classify vector-boson-fusion Higgs-pair events by coupling strength from particle physics simulations.
Machine LearningDeep LearningParticle Physics
2025 STYLO-Pipeline: Semantic Outfit Transformation
Felix Förster*, Andreas Weber*, Lars Schneider*
Vision-foundation-model pipeline for virtual try-on and semantic outfit editing — swap backgrounds, generate garments from text, and fit them onto a person with StableVITON.
Computer VisionFoundation ModelsGenerative AI
2025 Speeding Up pixelNeRF by Upsampling Low-Resolution Images
Felix Förster*, Lars Schneider*, Tobias Neumeier*, Adrian Struzek*
Cutting pixelNeRF's inference time ~19x by rendering downsampled images and recovering detail with a learned CNN upsampler, evaluated on SRN-Cars and CO3D-Apple.
Computer VisionDeep Learning3D Geometry
2024 Playing Card Object Detection
Felix Förster, Aaron Mahlke
Flutter companion app for the drinking card game that recognizes drawn cards on-device with a fine-tuned YOLO model and computes win-probability recommendations via Monte Carlo simulation of the remaining deck.
Computer VisionDeep LearningOn-Device ML
2024 Building Materials Segmentation with U-Nets
Felix Förster
A U-Net segmenting air voids, aggregate, and cement paste in concrete micrographs and predicting a relative height map, trained with active learning and synthetic image compositing.
Computer VisionDeep LearningSemantic Segmentation
Work Experience
Computer Vision Group (MuMoL), TUM
Apr 2025 — Present
Student Assistant · Munich, Germany
- Adapted open-source software for medical purposes, including local large-scale deployment of LLMs.
- Engineered LLM agents with LangGraph and improved performance through custom prompt engineering.
- Managed and analyzed datasets as part of ongoing research work.
- Reviewed third-party submissions for the IEEE Intelligent Vehicles Symposium (IV) 2025.
LLMs LangGraph Prompt Engineering Deployment
Levigo Solutions GmbH
Aug 2021 — Oct 2024
Working Student – Software Engineer · Munich, Germany
- Introduced software quality assurance measures for Kubernetes clusters with Testkube and Postman to improve code quality and reduce redundant tasks.
- Implemented a PDF/A verification tool for documents in Java as a back-end API.
Java Kubernetes REST API Testkube
Education
M.Sc. Computer Science
Oct 2024 — Present
Technical University of Munich (TUM) · Munich, Germany
- Grade: 1.3. Specializing in Computer Vision, Natural Language Processing, and Deep Learning for 3D Geometry.
- Built a deep learning pipeline combining open-source models including Stable Diffusion 2.5, SAM 2, and pose estimation (grade: 1.0).
B.Sc. Computer Science
Oct 2021 — Sep 2024
Technical University of Munich (TUM) · Munich, Germany
- Grade: 2.0. Specialized in Artificial Intelligence and Economics, with hands-on deep learning experience in Python, PyTorch, and Computer Vision.
- Bachelor's thesis on Multi-Agent Reinforcement Learning (grade: 1.0).
- Analyzed building materials with U-Nets and GANs (grade: 1.3).
Tech Stack
Machine Learning & Deep Learning
- Computer Vision
- Natural Language Processing
- Reinforcement Learning
- LLM Agents
- Audio Processing
- 3D Geometry
Programming & Tools
- Python
- PyTorch
- Slurm
- vLLM
- LangGraph
- Java
- C
- SQL
- Dart/Flutter
- Assembly
- Git
- Docker
- Kubernetes
- JUnit
Personal
Languages
- German (Native)
- English (C2)
- Spanish (B1)
Passions
- Snowboarding
- Running
- Mechanical Keyboards
- Gaming
Get in touch!
I'm always happy to hear about research, collaborations, or interesting problems in deep learning.