Work experience

Consultant lead engineer (May 2025 to date)

Assignment I

A concrete tile manufacturer required to automate visual quality inspection to improve product quality, reduce manual inspection effort, and increase consistency in defect detection across production shifts.

Problem

Existing manual inspection processes were subjective, prone to operator variability, and unable to reliably identify surface defects at production speed. The customer required a ML-driven quality control solution capable of detecting multiple defect types while integrating with industrial manufacturing processes.

Results

• Delivered a production-ready automated visual quality control system capable of detecting defects in concrete tiles with high accuracy and repeatability.

• Reduced dependency on manual inspections and improved inspection consistency across production shifts.

• Established a scalable machine learning framework that enabled future model retraining and continuous improvement.

• Successfully transferred knowledge and solutions to customer operational teams.

Tasks & Responsibilities

• Led a team responsible for the design and implementation of the complete computer vision and machine learning solution.

• Managed project planning, sprint execution, and technical decision-making using Agile methodologies.

• Defined the vision system architecture, including camera selection, illumination design, installation, calibration, and performance tuning.

• Established data collection and annotation workflows using CVAT to create high-quality training datasets.

• Designed and implemented computer vision pipelines for image acquisition, pre-processing, defect localization, segmentation, and classification.

• Developed and trained UNet segmentation models and YOLO-based object detection models using Ultralytics and PyTorch.

• Integrated MLflow for experiment tracking, model versioning, performance comparison, and deployment management.

• Led model validation and deployment activities within the industrial production environment.

Keywords

Team Leadership, Computer Vision, Machine Learning, MLOps, YOLO, UNet, PyTorch, CVAT, MLflow, Open3D, Point Cloud Processing, Industrial Quality Control, Agile Delivery.

Assignment II

A customer required a secure, self-hosted machine learning ecosystem to develop and deploy computer vision models for object detection within ship recycling environments.

Problem

The project required not only accurate object detection models but also a reliable ecosystem for data annotation, model experimentation, and deployment.

Results

• Delivered a robust MLOps foundation capable of supporting multiple machine learning use cases beyond the initial object detection project.

• Improved experiment traceability, model governance, and deployment readiness.

• Increased development efficiency through reusable architectures and automation.

• Enabled scalable management of datasets, experiments, and production models.

Tasks & Responsibilities

• Led the technical team responsible for computer vision and machine learning operations.

• Defined annotation workflows and quality assurance processes using CVAT.

• Selected and integrated MLflow for centralized experiment tracking and model management.

• Designed storage architecture using MinIO to support large-scale image datasets and machine learning artifacts.

• Established development standards for computer vision applications using PyTorch and Ultralytics.

• Created reusable model templates for training, evaluation, validation, and deployment.

• Coordinated Agile delivery activities including sprint planning, backlog refinement, and technical reviews.

• Mentored team members on MLOps best practices and machine learning engineering principles.

Keywords

Team Leadership, Machine Learning Architecture, Computer Vision, Self-Hosted Infrastructure, PyTorch, Ultralytics, MLflow, MinIO, CVAT, NetBird, Agile Delivery.

Assignment III

A customer required a robotic milling solution capable of accurately following wall geometries during automated machining operations. Precise wall alignment was required to ensure milling trajectories matched the actual physical environment.

Problem

Variations in installation tolerances, wall surface irregularities, and measurement noise introduced positioning errors that affected robot trajectory calculations and milling quality.

Results

• Delivered a robust wall alignment solution enabling accurate robot trajectory calculations.

• Improved milling precision by compensating for real-world installation deviations.

• Enabled reliable alignment between scanned environments and CAD references.

• Enhanced quality control through automated metrology verification.

Tasks & Responsibilities

• Designed and implemented computer vision algorithms for wall alignment and robotic trajectory correction.

• Developed point cloud acquisition and preprocessing pipelines to improve measurement quality and reduce sensor noise.

• Implemented geometric feature extraction techniques to identify reference planes, edges, and wall structures.

• Developed point cloud registration algorithms to align scanned environments with design references.

• Created obstacle detection algorithms to identify environmental objects that could interfere with robot operation.

• Worked closely with robotics and automation engineers to integrate vision-based alignment results into trajectory planning workflows.

• Validated algorithm performance through metrology-based accuracy assessments.

Keywords

Computer Vision, Point Cloud Processing, 3D Vision, Geometric Algorithms, Industrial Automation, 3D Metrology, Obstacle Detection, Part-to-CAD Registration.

Senior Research Engineer (Nov. 2019 to Feb. 2025)

Assignment

A medical technology company was developing vision care solutions to support cataract surgery procedures and improve the accuracy of ophthalmic measurements through augmented reality technologies.

Problem

The projects required the integration of hardware engineering, software development, computer vision, image processing, and clinical requirements into commercially viable products.

Results

• Delivered innovative vision care technologies combining computer vision, augmented reality, and medical imaging capabilities.

• Improved real-time visualization performance and product reliability.

• Improved engineering decision-making through data-driven analysis and experimentation.

• Delivered an innovative video-processing solution capable of visualizing otherwise difficult-to-observe ocular surface activity.

• Improved accessibility of physiological insights through enhanced visualization techniques.

• Improved alignment accuracy and stability of augmented reality visualizations.

• Reduced system latency through optimized software and hardware coordination.

• Contributed to the successful development of an advanced ophthalmic guidance platform supporting cataract surgery procedures.

• Enabled reliable augmented reality functionality within a medical device environment.

Tasks & Responsibilities

Research Engineer (Nov. 2019 to Dec. 2023)
Senior Research Engineer (Jan. 2024 to Feb. 2025)

• Collaborated with multidisciplinary teams including optics, hardware, software, and clinical specialists in the development of ophthalmic guidance technology.

• Contributed to hardware architecture decisions focusing on temperature stability, electronic reliability, and system performance.

• Performed hardware latency analysis and evaluated alternative hardware components to improve real-time responsiveness.

• Supported EMC certification activities and power delivery architecture design to ensure compliance and operational stability.

• Designed and implemented computer vision and image processing algorithms for dynamic augmented reality overlays.

• Developed calibration and registration algorithms to ensure accurate visualization and alignment of virtual information with real-world surgical views.

• Created design documentation, requirements specifications, and technical reports.

• Participated in Agile software development activities including sprint planning, technical reviews, and iterative product development.

• Developed prototypes and proof-of-concept solutions to validate technical feasibility.

• Defined data collection methodologies and data quality assurance procedures.

• Managed datasets and supported experimental validation activities.

• Defined software requirements and translated stakeholder needs into technical specifications.

• Conducted data analysis and experimental validation using Python, Pandas, Scikit-learn, and visualization frameworks.

• Developed software solutions using Python, OpenCV, PyTorch, NumPy, Pandas, Scikit-learn, FFMPEG, and Git-based workflows.

Keywords

Medical Devices, Ophthalmology, Surgical Guidance Systems, Augmented Reality, Computer Vision, Image Processing, Video Analytics, Research & Development, Product Innovation, Hardware-Software Integration, Real-Time Systems, Requirements Engineering, Clinical Technology,  Python, OpenCV, PyTorch, NumPy, Pandas, Scikit-learn, FFMPEG, Data Analysis, Agile Product Development.

Researcher (Oct. 2010 to Oct. 2019)

Assignment

Over several years, I served as a Principal Researcher across a multiple research initiatives involving healthcare technology, hyperspectral imaging, image fidelity assessment, telemedicine, texture analytics, and environmental sensing, while supporting collaboration between academia, research institutes, hospitals, and industrial organizations.

Problem

Each project addressed a different scientific and engineering challenge, including tissue hypo-oxygenation detection from hyperspectral imagery, micro-location air pollution prediction, skin cancer diagnostic imaging quality, image fidelity evaluation, telesurgery video quality assessment, and automated texture wear evaluation. The common challenge was developing objective, data-driven methodologies to characterize image and video content while ensuring scientific rigor and practical applicability.

Results

• Improved understanding of image fidelity, medical image quality, texture characterization, and video quality assessment.

• Improved objective assessment methodologies for image and video quality evaluation.

• Enabled the successful completion of multiple innovation projects funded by industrial and academic partners.

• Developed scientific expertise and mentored researchers contributing to long-term innovation and knowledge creation.

• Advanced methodologies in medical imaging, hyperspectral imaging, image quality assessment, video quality evaluation, texture analytics, and data-driven computer vision.

Tasks & Responsibilities

Research assistant (Oct. 2010 to Jan. 2012)
Researcher (Sep. 2012 to Oct. 2019)
Coach of junior PhD students (Nov. 2018 to Oct. 2019)

• Led industry-university collaborative research projects and managed technical, scientific, and operational activities.

• Supervised junior PhD researchers and provided guidance in computer vision, image processing, experimental design, and scientific validation.

• Managed project planning, stakeholder alignment, scheduling, deliverables, reporting, and multidisciplinary collaboration.

• Developed image registration and image analysis algorithms for hyperspectral imaging and tissue hypo-oxygenation detection.

• Directed activities related to environmental monitoring and air pollution prediction systems.

• Evaluated image quality characteristics and image fidelity measures for dermatological imaging applications.

• Evaluated compressed video quality for telesurgery applications using objective video quality assessment methodologies.

• Investigated texture analysis, feature engineering, and algorithm benchmarking for wear characterization applications.

• Authored technical reports, research documentation, and project deliverables.

• Developed content-aware image fidelity evaluation approaches addressing color consistency, color difference measurement, and contrast preservation challenges.

• Investigated image quality assessment methods for interventional X-ray applications.

• Conducted statistical analysis, scientific experimentation, performance benchmarking, and technical reporting using Python, OpenCV, Scikit-learn, Pandas, NumPy, and Matplotlib.

Keywords

Principal Researcher, Industry-University Collaboration, PhD Coaching, Medical Imaging, Hyperspectral Imaging, Computer Vision, Image Quality Assessment, Video Quality Assessment, Texture Analysis, Environmental Monitoring, Statistical Analysis, Data Science, Scientific Innovation, Image Processing, Fidelity Assessment, Video Compression, Applied Research, Image Analytics, Video Analytics, Machine Learning,  Python, OpenCV, PyTorch, NumPy, Pandas, Scikit-learn.