Hello, I'm

Max Keller

Senior Robotics & Machine Learning Engineer

Building intelligent robotic systems with cutting-edge ML. Specializing in Vision-Language Models, autonomous driving, and end-to-end learning systems. 12+ patent applications filed. Based in Berlin.

View Projects Get in Touch

About Me

Bridging Research and Production in Robotics & AI

Passionate about Intelligent Systems

I'm a robotics and machine learning engineer with expertise in developing end-to-end autonomous systems. My work spans from large-scale multimodal data pipelines and Vision-Language Model training to real-time deployment.

At Sensmore, I design systems that process 100+ TB of robotics data and deploy VLMs on Jetson hardware. Previously at Bosch Research, I developed and validated hybrid planning algorithms for automated driving, contributing to 12+ patent applications.

I hold a M.Sc. in Computer Science (Intelligent Systems) and a B.Sc. in Physics, both from University of Stuttgart.

12+
Patent Applications
100+
TB Data Processed
25
Invention Disclosures
6+
Years Experience
1.3
M.Sc. CGPA
2
Publications

Experience

Building the Future of Autonomous Systems

February 2025 - Present

Senior Robotics & Machine Learning Engineer

Sensmore, Berlin
  • Designed continuous multimodal data pipelines processing 100+ TB of robotics data for VLM training
  • Built self-supervised scenario mining pipelines using SAM3, Florence-2, and Gemini VLMs, accelerating discovery by 10x
  • Automated ~90% of VLA/VLM data generation, significantly reducing manual labeling cost
  • Fine-tuned the first in-house 7B VLM using DeepSpeed and Hugging Face
  • Deployed VLM on Jetson Thor for real-time inference within ROS2-based robotics system
  • Developed system identification framework and driving control stack for autonomous operation
January 2023 - December 2024

Research & Software Engineer

Robert Bosch GmbH (Research), Renningen
  • Researched and deployed model- and data-driven planning algorithms for automated driving
  • Developed Bosch's first hybrid planner, achieving 15% performance improvement in real-vehicle testing
  • Led vehicle integration and ROS2-based architecture development across multiple platforms
  • Co-authored 25 invention disclosures (10+ first author), resulting in 12+ patent applications
  • Supervised and mentored a Master's thesis student through successful completion
April 2022 - December 2022

Intern & Master Thesis Student

Robert Bosch GmbH (Research), Renningen
  • Implemented deep-learning research papers with 13% improvement over local baseline
  • Developed novel trajectory prediction approach using future context (10% improvement)
  • Reduced technical debt of main code-base by 20%
  • Submitted two patent applications and wrote one successful research publication
February 2019 - March 2022

Working Student (Software Architecture)

Robert Bosch GmbH, Leonberg
  • Developed tool for automated C++ code analysis, found 100+ missing ports in software architecture
  • Integrated modules into the architecture and main codebase
  • Onboarded and mentored three new working students
October 2020 - April 2021

Research Assistant

University of Stuttgart
  • Wrote transpiler to convert Qiskit circuits into MW/RF waveforms for 3-qubit NV-Center prototype
  • Developed web interface for remote execution of Jupyter Notebooks on quantum hardware

Publications

Peer-Reviewed Research Contributions

IEEE IV 2023

Bridging the Gap Between Multi-Step and One-Shot Trajectory Prediction via Self-Supervision

Max Keller et al.

View on arXiv
arXiv Preprint 2022

On Readout and Initialisation Fidelity by Finite Demolition Single Shot Readout

Max Keller et al.

View on arXiv

Featured Projects

Open Source Contributions & Personal Projects

MeetChrist

Cross-platform Christian social media app featuring content sharing, event management, user networking, and real-time group chat functionality.

Flutter Firebase Docker
View Repository

Deep Active Learning Framework

Open source contribution to modAL (2,193+ stars). Implemented Monte-Carlo dropout based deep active learning methods for efficient model training.

Python PyTorch modAL
View Repository

CNN Beam Hardening Artifact Reduction

Deep learning solution for CT scan artifact reduction. Trained CNN models, generated datasets, and developed volume data segmentation tools.

Python PyTorch C++
View Repository

Technical Skills

Technologies & Tools I Work With

Programming Languages

Python C++ C Java TypeScript Angular

ML & Robotics

PyTorch HuggingFace DeepSpeed ROS1/2 OpenCV TensorRT ONNX CUDA Ray Scikit-learn

Computer Vision & VLMs

Vision-Language Models SAM Florence-2 Gemini Trajectory Prediction Object Detection

Autonomous Systems

Motion Planning MPC System Identification Control Systems Sensor Fusion

DevOps & Infrastructure

Linux Git Docker Jenkins GCP SQL

Project Management

Scrum JIRA Confluence Linear Notion

Education

Academic Foundation

M.Sc. Computer Science

University of Stuttgart, Germany
October 2020 - December 2022
CGPA: 1.3 (Excellent)
  • Specialization: Intelligent Systems
  • Deutschlandstipendium awarded for outstanding achievements
  • Key courses: Computer Vision, Deep Learning, Math for Intelligent Systems, Software Architecture

B.Sc. Physics

University of Stuttgart, Germany
September 2017 - September 2020
CGPA: 1.5 (4th/54 graduates)
  • Thesis: Classical feedback control of a 3-qubit system
  • Research results published in peer-reviewed journal
  • Key courses: Quantum Information Processing, Computational Physics, Statistical Mechanics

Get In Touch

Let's Build Something Amazing Together

I'm always interested in discussing new opportunities in robotics, machine learning, and autonomous systems. Feel free to reach out!

max.keller2312@gmail.com LinkedIn GitHub