Welcome to My Professional Profile
Researcher | Academic | Professional
I am a Senior Technical Leader in Automated Driving Groundtruth Systems and Data-Driven ML Platforms. My research areas span LLM fine-tuning and interpretation, deep learning, sensor fusion, and auto-labelling systems for multi-modal sensor data
Curriculum Vitae
Professional background and academic credentials
Professional Experience
Senior Technical Lead – AI & Data Platform Development
Robert Bosch GmbH - Automated Driving June 2022 - Now- Drive cross-cluster alignment across the end-to-end data and labelling value chain
- Operate in a de-facto senior leadership role, shaping strategy, governance, and execution
- Define and execute technical and organizational roadmaps across multiple teams
- Establish KPI frameworks and data-driven steering mechanisms
- Contribute to organizational design, leadership enablement, and scaling of engineering units
- Define and evolve architecture for large-scale cloud-based data platforms
- Ensure consistency and scalability across a multi-team ecosystem
- Drive platform standardization and reduce redundancy across teams
- Lead technical decision-making for data-intensive, ML-driven systems
- Improved end-to-end data flow and delivery reliability by resolving systemic bottlenecks
- Enabled scalable platform evolution for high-volume sensor data and labelling workflows
Technical Lead – Data Selection (Video Perception)
Robert Bosch GmbH - Automated Driving June 2022 - Now- Led an international team of up to 19 engineers
- Owned roadmap planning and stakeholder alignment across product and engineering
- Designed and built a scalable cloud-based data selection platform for video perception workloads
- Defined system architecture for large-scale data processing and maintainability
- Worked on ML-driven data selection and pipeline optimization
- Delivered production-ready platform enabling efficient processing of large-scale sensor data
Deep Learning Engineer
Robert Bosch GmbH - Automated Driving April 2021 - June 2022- Developed ML models for perception and sensor data processing
- Contributed to ML pipelines and tooling for large-scale datasets
- Took on early coordination and technical leadership responsibilities
PhD Program
Robert Bosch GmbH - Automated Driving April 2017 - April 2021- Developed ML-based approaches for sensor fusion and data labelling
- Published in IEEE and Nature Scientific Reports; multiple conference contributions
- Contributed to patent applications and supervised students
- Bridged academic research with production-relevant data systems
Research Areas
Current and past research focus areas
LLM finetuning and interpretation
My research focuses on efficient fine-tuning of LLMs using parameter-efficient methods like LoRA/QLoRA (implemented for Mistral-7B), rigorous evaluation via modular benchmarking suites (MMLU, TruthfulQA, ARC), and training data analysis to understand how dataset composition, quality, and bias impact model behavior. This holistic approach—spanning training, evaluation, and data-centric insights—enables development of more reliable, interpretable, and capable language models. My work emphasizes reproducibility, practical deployment, and the critical interplay between data, architecture, and assessment methodologies.
Learn MoreAuto-labelling & Data Selection
My professional work and research focuses on scalable auto-labelling and efficient data selection for multi-modal data. I have extensive experience working with sensor data (e.g., perception systems), which includes sensor-data, time-series and structured/semi-structured data, and have applied AI methods to improve data efficiency, system performance, and scalability.
Learn MoreMulti-modal sensor fusion and semi-automated labelling systems
My PhD research focused on applied deep learning for multi-modal sensor data, particularly in the context of automated driving. I developed methods for sensor fusion and semi-automated data labelling, bridging machine learning research with scalable, production-relevant systems. My work emphasized efficiency and practicality in handling large, complex datasets.
Learn MoreRecent Projects
Selected research projects and collaborations
Autolabelling Methods and Pipelines
06/2022 - nowDevelopment of methods for and bring-up of a large scale autolabelling pipeline - with Cariad and Bosch Automated Driving.
View ProjectData Selection Methods and Pipelines
06/2022 - nowTargetted data selection on scale in cloud and in vehicle - with Cariad, Bosch Research and Bosch Automated Driving.
View ProjectMulti-modal Object Detection
04/2017 - 04/2021Developing and Analysing methods for multi-sensor object detection - with Bosch Research and Bosch Automated Driving.
View ProjectPublications
Peer-reviewed publications and research outputs
Deep Multi-modal Object Detection and Semantic Segmentation for Autonomous Driving: Datasets, Methods, and Challenges
IEEE Transactions on Intelligent Transportation Systems, 22(3), pp. 1341-1360
This review paper systematically summarizes methodologies and discusses challenges for deep multi-modal object detection and semantic segmentation in autonomous driving, providing an overview of on-board sensors, open datasets, and background information for object detection and semantic segmentation in autonomous driving research. We then summarize the fusion methodologies and discuss challenges and open questions.
Adaptive Stochastic Resonance for Unknown and Variable Input Signals
Scientific Reports, 7(1), 2450
All sensors have a threshold, defined by the smallest signal amplitude that can be detected. The detection of sub-threshold signals, however, is possible by using the principle of stochastic resonance, where noise is added to the input signal so that it randomly exceeds the sensor threshold. The choice of an optimal noise level that maximizes the mutual information between sensor input and output, however, requires knowledge of the input signal, which is not available in most practical applications. Here we demonstrate that the autocorrelation of the sensor output alone is sufficient to find this optimal noise level. Furthermore, we demonstrate numerically and analytically the equivalence of the traditional mutual information approach and our autocorrelation approach for a range of model systems. We furthermore show how the level of added noise can be continuously adapted even to highly variable, unknown input signals via a feedback loop. Finally, we present evidence that adaptive stochastic resonance based on the autocorrelation of the sensor output may be a fundamental principle in neuronal systems.