CTO and Director at Corpy&Co. with expertise in AI/ML, mathematical modeling, and technical leadership. Passionate about transforming complex research into practical solutions to real-world problems.
I am an accomplished technology leader with a Ph.D. in Computational Geometry. Currently serving as CTO and Director at Corpy&Co. in Tokyo, Japan, where I lead a team that has grown from 10 to 70 engineers during my tenure.
During my five and a half years at Corpy, my team and I have successfully delivered over 50 client-facing R&D projects. I specialize in translating complex research into practical applications while managing cross-cultural teams and high-profile client relationships. In parallel, I've led my team to build CONFIDE, a cutting-edge AIOps application to enable adoption of mission-critical AI in manufacturing environments.
During the past few years, my interest has naturally shifted to the broad topic of Responsible AI, which covers Explainable AI and Quality-Assurable AI, as well as AI Ethics, AI Safety, and other related fields. My interest was developed initially due to the focus of CONFIDE, however I have come to find personal interest in these areas, as I firmly believe that such issues are critical in today's world where AI is evolving rapidly and there are so many initiatives for its adoption in multiple fields. My opinion is that such adoption must be accompanied by appropriate measures, including ethical concerns, regulation, and legislation.
My academic background is in Applied Mathematics and Computational Geometry. I find excitement in combining seemingly unrelated fields or ideas to create new solutions to real problems. During my PhD, I contributed to the CGAL, the largest open-source library for computational geometry. The algorithm developed during my doctoral research is now part of the library.
Expertise in building and deploying AI-based systems for Mobility, Manufacturing, and Healthcare. Strong interest and proven experience with Quality Assurance for AI.
Expertise in design and analysis of algorithms for academia and industry. Long track record of delivering high-quality software solutions.
Grown engineering team from 10 to 70 members. Brought together people from different backgrounds to build an inclusive and supportive environment.
Multilingual professional experienced in cross-cultural environments across Europe and Asia. Experience in balancing business, technical, and ethical requirements.
Key skills and technologies I've developed throughout my professional journey
Professional journey: technical and leadership roles in AI R&D, as well as software engineering in four countries
Expanded my responsibilities at Corpy to include higher management and strategic decision making, growing the Engineering team to 70 members while continuing to oversee technical R&D initiatives.
Led technical R&D efforts across multiple AI/ML projects, while simultaneously establishing the foundations for a cutting-edge AIOps platform.
Developed a game streaming client applications for Samsung TVs using proprietary SDK.
Conducted applied research in Computational Geometry, resulting in one package contributed to the CGAL library.
Applied geometric processing to medical research focused on abdominal aortic aneurysms.
Developed software solutions for business applications using InstantDeveloper.
Developed applications for Windows and Windows CE devices.
Strong foundation in applied mathematics, computer science, and computational geometry
Thesis: Delaunay triangulations of a family of symmetric hyperbolic surfaces in practice
Researched the properties of triangulations of the Bolza surface. Formulated and implemented algorithms for the computations of such periodic triangulations. The implementation of my research is now part of the CGAL library, the largest open-source library for computational geometry.
Thesis: Shape-Preserving Interpolation on the Sphere
Implemented and analyzed shape-preserving interpolation techniques on the unit sphere. Developed a novel approach to find optimal tension parameters that preserve the shape defined by control points on a sphere. The code is available on GitHub.
Thesis: The Euclidean InSphere Predicate
Developed a robust method to compute the InSphere predicate in 3D by computing the sign of expressions of algebraic degree at most 10. Previous methods relied on expressions of degree at most 14, so my work was a significant improvement of the state of the art.
A small sample of projects in which I was directly involved as developer or researcher
Disclaimer: The AI/ML projects highlighted in this section are arbitrarily chosen among the 50+ client-facing projects I have supervised or in which I participated during my work at Corpy&Co., Inc. All information shared here is intentionally limited to avoid disclosing specific technical implementations, proprietary methodologies, and confidential client information.
CONFIDE is an AIOps platform developed to facilitate the adoption of AI solutions in mission-critical contexts in the manufacturing industry. Composed of multiple modules, CONFIDE offers the tools to manage, annotate, generate, and evaluate the quality of image data (text and tabular data are on the roadmap). Once the data fulfils business requirements, AI models can be trained and compared in order to identify the one that fits best the targets of the company. The model can be evaluated and compared with other models in a transparent manner, with meaningful metrics, directly addressing the main barriers to AI adoption in manufacturing: the lack of understanding and trust in "black box" AI systems.
The platform implements rigorous quality assurance methods implemented by leading research institutes in Japan, enabling manufacturing teams to confidently integrate AI into critical processes while maintaining compliance with quality standards. CONFIDE is currently being introduced to major clients and is in the process of adapting to fit their business needs.
Fundamental research on Quality Assurance for black-box and white-box LLMs. Investigated the robustness of two LLMs to various transformations of the input prompt, and presented preliminary principles for the formulation of Quality Assurance guidelines.
Quality Assurance for Object Detection systems in the context of Autonomous Vehicles. Considering the SOTIF and PEGASUS principles, implemented a comparative study of different Object Detection models to demonstrate their suitability for ADAS applications.
Implementation and validation of a custom method for the detection of unknown (i.e., out-of-distribution) objects with the use of a one-step object detector. The method demonstrated state-of-the-art performance in the domain and is being used for the detection of obstacles.
Developed and implemented an algorithm for the optimization of the manufacturing process of industrial batteries. The method combines GAN (Generative Adversarial Networks) and Genetic Algorithms to explore new combinations while maintaining feasibility constraints.
Supervised and contributed to the implementation of a system for the monitoring and detection of calving events. The AI detection model is integrated into a real-time notification and alerting system to ensure the safety of calves during the calving process.
Developed and implemented an algorithm for constructing periodic Delaunay triangulations of the Bolza surface. The resulting code is now part of the CGAL library, the largest open-source library for computational geometry.
Developed geometry-based methods to determine morphological changes in abdominal aortic aneurysms between scans. This work provided valuable diagnostic insights for medical professionals to guide treatment of patients.
Implemented a shape-preserving interpolation method on the unit sphere. The implementation is provided in C++ and includes visualization scripts. The proposed algorithm finds optimal tension parameters to ensure geometric continuity at defined control points.
Beyond professional accomplishments and technical projects -- additional info about me and my interests
I often get emails addressed to "Dear Lordan" -- it's actually "Iordan" (starting with an uppercase 'i'). It gives me a chuckle every time, and it's always a pleasant surprise when people get it right!
I was born in Bulgaria and have Bulgarian nationality, but I also belong to the nomadic Boyash tribe (also known as Ludari, Rudari, or Linguri). We have our own language derived from Latin with elements of Bulgarian. Interestingly, I've met Linguri in Greece, Italy, and France, despite geographical separation.
I speak five languages fluently: English, Greek, Italian, French, and Bulgarian. Additionally, I speak my native Boyash language, but that is not an "official" language. Despite living in Japan for almost six years, my Japanese is still not good.
For those who enjoy mathematical and programming challenges, I recommend Project Euler, where you can find a diverse set of problems combining mathematics and programming. It's an excellent way to develop problem-solving skills and explore interesting mathematical concepts.
John Horton Conway's Game of Life demonstrates how complex patterns can emerge from simple rules. I implemented the Game of Life in JavaScript on a periodic space (flat torus), allowing patterns to wrap around the edges. It's quite interesting!
An interactive animation based on Game Theory that demonstrates how trust evolves in different scenarios. It provides fascinating insights into cooperation and competition.
MIT's Moral Machine presents modern versions of the trolley problem in the context of autonomous vehicles. It raises important ethical questions about AI decision-making in life-or-death situations.
I'm open to discussing professional opportunities, research collaborations, or any questions related to my work and expertise