Thierry Backes

- Evaluation of intrinsic camera calibration and pose estimation pipeline - SLAM and LiDAR inertial navigation

- Linux realtime scheduling optimization and introduction of low-level tracing framework - Leading LiDAR selection and evaluation process - Introducing a new local development environment - Introducing new team processes from documentation to issue triaging workflows

- Extrinsic multi-view camera calibration - Intrinsic camera calibration C++, ROS(2), Python, OpenCV, PyTorch, PCL

- Camera based obstacle detection, tracking, and collision prediction - LiDAR based forward looking obstacle detection and collision prediction - LiDAR camera calibration - Sensor integration and driver development C++, ROS, Python, OpenCV, PyTorch, PCL

- Conducting interviews for all technical positions - Onboarding new members - Guiding team to become independent within 6 months - Reviewing technical work

Core member of AMZ Driverless perception team. The goal is to create a driverless racing car from scratch and compete in formula student competitions. My main work is in object detection with deep learning and sensor fusion between Lidar and camera setup. Data analysis is done on the different perception pipelines to get statistical errors which are required for probabilistic location estimation. In addition, the team explores monocular depth estimation via deep learning and pose estimation, as well as object and color detection with Lidar data. A photorealistic simulator is being developed to have faster iteration times on different sensor setups and pipeline changes. Part time commitment the first semester, full-time the second. Software stack includes C++, Python, Scikit and Pytorch and Tensorflow, ROS, Linux and Git, as well as a CICD setup with Jenkins. I also help out on the electric vehicle components which include the high- and low-voltage system as well as the electronic control unit which communicates via CAN bus. Sensor setup are 3 RGB global shutter cameras and 2 Hesai Lidars.

Worked on mobile data analysis and delivered multiple client projects. Big data analysis done with Scala and Spark. Also worked on internal tools development and prototyping with Scala, Python, Docker, Linux, PostgreSQL, and Clickhouse. Timeseries analysis done in Python with scikitlearn and pandas. Deployment with Ansible and Kubernetes on bare-metal and GKE. Used QGIS and Mapshaper for GIS. Data visualization with Tableau. Additionally used Git and CircleCI.

Software development and DevOps. Integration and maintenance of CI/CD system to existing infrastructure and development of internal tool chain for software project creation, update and management. Stack: Python 3, GRPC, Gnu/Linux, Ansible, Teamcity (CI/CD), Docker, LDAP, OAuth, Git.

Worked full-time during the summer break as a 2D Unity developer for Project Rosie targeting mobile applications. Focus were educational games for children to learn science with the help of the character "Robot Rosie". Children had to use their phone sensors to explore the world. Worked with Unity and C#, as well as Git.

Focus on Datascience, Machine Learning, Deep Learning, and Statistics. Master Thesis is about kinetic parameter estimation for metabolic networks in computational systems biology with (generalized) linear mixed models.

Bachelor Thesis was about "Asynchronous Data-Centric Raytracing"