Dimitrios Dagdilelis
ML/AI Engineer
Biography
AI/ML Engineer with a PhD in multi-modal data fusion. I empower organizations and teams to navigate their AI transformation journeys. By delivering impactful, high-value data products, I ensure that business strategies are backed by real, scalable AI solutions while separating genuine advancements from overhyped, unreliable claims. My focus is on delivering robust, production-ready systems that solve real-world problems and drive sustainable growth.
Interests
- Multimodal AI
- Computer Vision & LLMs
- MLOps
- Scalable AI
- Explainable AI
Education
PhD in Multi-Modal Data Fusion, 2024
Technical University of Denmark
MSc Automation & Robot Technology, 2020
Technical University of Denmark
Diploma in Electrical & Computer Engineering, 2016
Aristotle University of Thessaloniki
Experience
Senior AI/MLOps Engineer
Danish Defence
- Built strong baselines and reproducible evaluation pipelines aligned with current SOTA practice for deep learning model development.
- Designed and deployed a high throughput computer vision stack covering image classification, instance level retrieval, and 3D scene parsing using modern transformer based and metric learning architectures.
- Developed model introspection and attribution tools to verify behavior, detect failure modes, and validate predictions under real world conditions.
- Implemented a production grade visual search system for large scale image retrieval over indexed multimodal datasets using dense and hybrid embeddings.
- Created automated labeling and active learning loops that reduced manual annotation load and increased data efficiency across training cycles.
- Built robustness and data quality diagnostics to measure distribution shift, detect data drift, and track reliability of deployed models in continuous operation.
- Established concrete governance standards for experiment tracking, reproducibility, and model documentation, and led technical sessions to ensure alignment with contemporary research practice.
AI R&D Engineer
- Led the development of the perception stack of the first autonomous ferry in Denmark, delivering a production-grade situational awareness system (a-look-inside) (demo-video).
- Designed a multimodal ML fusion engine, enabling 3D scene understanding and target tracking, combining radar, LiDAR, and multi-view camera data (preview).
- Scaled data operations through automated and semi-automated annotation pipelines, dataset versioning, and visual data exploration tools (preview).
- Partnered with downstream product owners to triage failure modes, collect targeted in-field & synthetic data, deploying iterative model improvements.
- Worked extensively with 3D perception and geometric reasoning tasks, enabling 3D object detection, tracking, and surrounding environment segmentation.
Visiting AI Engineer
- Extended an existing 2D detection pipeline into a 3D object detection system for maritime navigation (preview).
- Curated large datasets and built visualization tools to accelerate model iteration and evaluation.
- Collaborated effectively in hybrid remote/on-site settings, aligning research milestones with product delivery.
PhD
Technical University of Denmark (DTU)
- Research focused on multimodal data fusion, deep learning, and 3D computer vision.
- Thesis available here.
