It's Karim Janou
A AI Software Engineer
AI / Computer Vision Engineer building perception models for automotive and embedded systems. From training to deployment.
Innovation through code, passion in every line.
Contact Me
Contact Me
If things are still under control, then you are not going fast enough.
About me
AI / Computer Vision Engineer with ~4 years of experience developing and deploying perception models for automotive and embedded systems. Strong background in multimodal sensing (Camera, LiDAR, Radar), model optimization, and production-ready ML pipelines in industrial environments.
My Educational background.
Technische Universität München
M.Sc. Informatics
04/2020 - 04/2022
Lebanese American University
B.E. In Computer Engineering (Distinction)
09/2014 - 01/2020
0 +
Years Experience
0 +
Completed Projects
0 +
Companies
PyTorch
TensorFlow
LangChain
HuggingFace
Docker
AWS
NextJS
TypeScript
Work Experience
From research to production — building AI systems that drive real-world impact.
Here's a look at my professional journey so far.
11/2024 - Present
Computer Vision Perception Engineer
Aptiv (through Technology and Strategy)
Munich, Germany (Remote)
- ▹Optimized interior sensing models, achieving 10× inference speed-up via pruning and quantization.
- ▹Developed internal tooling to accelerate model deployment and debugging on target embedded hardware, reducing iteration time from ~1 day to ~30 minutes (≈95% faster turnaround).
- ▹Designed a quantization optimization tool reducing quantization error by 20–40%, significantly improving post-quantization model stability and performance.
10/2022 - 11/2024
Computer Vision Perception Engineer
Bosch (through Technology and Strategy)
Munich, Germany (Remote)
- ▹Developed 3D Detection/Tracking evaluation tool to generate customer-relevant KPIs. (NuScenes).
- ▹Trained and optimized 2D/3D object detection and sensor-fusion models (PV-RCNN, Detectron2, OpenPCDet, Vision Transformers) using DistributedDataParallel (DDP) multi-GPU training, improving detection performance by +50% mAP in internal benchmarks.
- ▹Developed and fine-tuned tracking algorithms (Kalman Filter), improving IDF1, ID switches, and fragmentation metrics by +50%.
- ▹Established requirements for the Cloud/Testing teams to deploy CI/CD pipelines on Azure.
05/2022 - 10/2022
Artificial Intelligence Associate
PricewaterhouseCoopers (PwC)
Munich, Germany
- ▹Consulted an L4 Autonomous Driving Startup on integrating radars in their perception pipeline and developing radar-based ego velocity compensation algorithm. (ROS2).
- ▹Built a proof-of-concept Question-Answering AI (Dense Passage Retrieval Transformer-Based Model) to support auditing tasks ( PyTorch, HuggingFace).
- ▹Contributed to the development of a methodology for AI platform certification under the EU AI act.
08/2020 - 05/2022
Software Testing Engineer - Working Student
Infineon Technologies
Munich, Germany
- ▹Automated testing for machine learning surveillance applications using Pytest.
- ▹Implemented CI/CD pipelines (Jenkins) and integrated data versioning tools (Jenkins, DVC, MLFlow).
- ▹Master's Thesis: Transferability of adversarial attacks between radar-based classification and regression neural networks (PyTorch, Foolbox) and generating universal adversarial perturbations using recurrent autoencoders.
04/2019 - 08/2020
Machine Learning Engineer (NLP) - Intern
BMW Group
Munich, Germany
- ▹Started as Intern (04/2019 – 12/2019), continued as Working Student (04/2020 – 08/2020).
- ▹Enhanced ontology matching for the OAEI conference (Python, NLTK, Gensim).
- ▹Developed Email based Error Notification System (Python, Redis Queues, Flask).
- ▹Full-stack development and maintenance of IT Helpdesk Chatbot using IBM Watson (Angular, .NET, Python, Docker, Traefik, Redis) .
- ▹Built a Chatbot Dashboard to monitor and analyze user interactions (Flask).
Featured Projects
Data is the soil, algorithms are the seeds.
A selection of side projects spanning computer vision, NLP, and GenAI.
Adventour - AI Travel Planner
A web application documenting global travel locations and helping create travel itineraries using AI.
Who's That Pokémon?
Lab Course: Face Verification and Retrieval (Siamese Networks)
Master Thesis
Transferability of adversarial attacks between radar-based classification and regression neural networks, and generating universal adversarial perturbations using recurrent autoencoders.
Sentiment Analysis
Performed sentiment analysis on tweets using BERT transformer model.
Explore more projects in my github profile
Want to collaborate?
Contact me!
Karim Janou
AI Engineer, Developer & Explorer
I'm always excited to cooperate on new projects and exchange valuable ideas that push the boundaries of AI and technology. Whether you have an innovative concept to explore, a challenge to solve together, or simply want to connect — feel free to reach out!