Andrei Costinescu
Scientific Researcher at Munich Institute of Robotics and Machine Intelligence (MIRMI)
Garching bei München, Bavaria, Germany
Summary
👤
Senior🎓
Top SchoolAndrei Costinescu is a scientific researcher and Ph.D. student at TUM/MIRMI with nine years of experience applying AI, machine learning, computer vision and humanoid robotics to assistive Geriatronics. He holds an M.Sc. in Informatics (grade 1.0) from TUM and spent an exchange term at Georgia Tech focusing on robot learning and mobile manipulation. His work spans scene context understanding from 6D object poses using graph-matching techniques and deep CNN-RNN models for surgical video analysis, bridging symbolic graph methods with deep learning. Andrei contributes to open-source projects as a technical writer (notably improving documentation for the popular Anki Cozmo Python SDK), reflecting attention to usability and developer experience. Based in Garching near Munich, he combines rigorous academic research with practical code and documentation improvements. Outside research he brings creative and disciplined dimensions to his work—former actor, guitarist and Taekwon-Do green belt.
9 years of coding experience
Master of Science - MS, Informatics, 1.0, Master of Science - MS, Informatics, 1.0 at Technical University MunichExchange student, Computer Science, A, Exchange student, Computer Science, A at Georgia Institute of TechnologyAbitur, Abitur at Colegiul German Goethe
German, English, French, Romanian
Stackoverflow
20reputation
435reached
3answers
1question
Github Skills (7)
documentation10
neural-network6
deep-learning6
tensorflow6
python6
lasagne6
theano6
Programming languages (5)
Github contributions (5)
anki/cozmo-python-sdk
Dec 2017 - Dec 2017
Anki Cozmo Python SDK
Role in this project:
Technical Writer
Technical WriterContributions:8 commits, 8 PRs, 7 comments in 1 day
Contributions summary:Andrei primarily contributed to improving the documentation of the Anki Cozmo Python SDK. Their commits involved correcting typos in docstrings and clarifying descriptions related to the Cozmo's functionalities, including the LiftPlate, animation filters, annotation priorities, and object/face visibility timeouts. They also corrected pip installation instructions in the README.md, ensuring clarity and accuracy in the SDK's documentation for users.
pythoncozmosdksdk-pythonanki
AndreiCostinescu/darknet
Aug 2020 - Dec 2024
YOLOv4 - Neural Networks for Object Detection (Windows and Linux version of Darknet )
Contributions:47 pushes, 4 branches, 6 tags in 4 years 5 months
pytorchwindowsdeep-learninglinuxobject-detection
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.