Vitae

Posdoctoral Researcher (3D Computer Vision)

SnT • University of Luxembourg (Luxembourg)

• Member of the Computer Vision, Imaging & Machine Intelligence Research Group (CVI²) at SnT, University of Luxembourg.
• My research focus is on reverse engineering of CAD models from 3D scans and CAD design generation (Text2CAD).
• Working in close collaboration with our industrial partner Artec3D.
• Group Head: Dr. Djamila Aouada
• Projects
  • MORFIS, is an AI design platform that empowers users, regardless of their technical background, to create, customize, and prototype physical products. [Project Page]
  • Deep Learning of 3D Scanned Data [Project Page]
  • CASCADES, Constrained Sequence modelling of CAD for reverse Engineering from 3d Scans [Project Page]
  • FREE-3D, Feature Reverse Engineering Of 3D Scans [Project Page]
• Research areas
  • VLLMs for CAD Design
  • Text2CAD
  • Reverse Engineering from 3D Scans
  • Reverse Engineering of Parametric CAD Sketches

Senior Machine Learning Engineer (Recommender Systems)

Taboola • DeepLab (Athens, Greece)

• Part of DeepLab's dedicated team of Taboolars.
• I aqured experience with developing and deploying ML solutions at scale, for Taboola's massive recommender system service.
• Our work on incremental learning for efficient Click-Through-Rate (CTR) prediction, was presented on the industrial track of [RecSys2022].
• I supervised two bachelor students and one master student along the lines of DeepLab's collaboration with the National Technical University of Athens (NTUA).
• Research areas
  • Deep Learning for Recommender Systems
  • Click-Through-Rate (CTR) prediction

Research Intern (Computer Vision)

Samsung AI Centre Cambridge (SAIC-C) (Cambridge, UK)

• Six-month PhD Internshp, extended to a part-time position.
• Manager: Dr. Yiorgos Tzimiropoulos
• I developed a framework for adapting pretrained pose estimation models on new target domains. Significant performance increase was reported on internal company benchmarks, with direct application on various pipelines within Samsung.
• Research areas
  • Multiperson Human Pose Estimation
  • 3D Pose Estimation
  • Semi-Supervised Learning
  • Self-Training

Software Engineer

Singular Logic (Athens, Greece)

• Designing custom solution, applications and services for organisations of the Greek public sector.
• Main focus on server-side applications, middle-wares and APIs, using .NET and C#.
• Technologies
  • C#, .NET Core, ASP.NET Web Api 2, ASP.NET Core, LINQ, Ninject, TFS, ADO.NET, Async Programming.

PhD in Computer Vision

University of Nottingham (Nottingham, UK)

• Advisor: Dr. Yiorgos Tzimiropoulos
• Member of the Computer Vision Laboratory (CVL) of the University of Nottingham.
• My PhD thesis was on Unsupervised Landmark Discovery via Self-Training Correspondence.
• Research conducted for my thesis has been partially published at [NeurIPS2020] and [TPAMI2023].
• Research areas
  • Unsupervised Landmark Discovery
  • Self-Supervised Learning
  • Self-Training
  • Learning under label noise

Diploma in Electrical and Computer Engineering

Democritus University of Thrace (D.U.Th) (Xanthi, Greece)

• Program Duration: 5 years
• A diploma is the Greek equivelant to BSc + MSc.
• My diploma thesis was on Sound Source Separation in Real Room Environment.
• Research conducted for my thesis has been partially published in [AIAI2016] and later [EvolvingSystems2018].
• Our work on sentiment analysis of Greek Tweets lead to publications in [PCI2015] and [JSIT2018].
• Graduated with the 4th highest grade on my cohord.
• Advisor: Dr. Nikos Mitianoudis

Master/Undergrad students (collaboration with Deeplab).

• Through my ongoing collaboration with Deeplab, I participate in the supervision master/undergrad student's, towards either the completion of their thesis or paper submissions.
• Part of Deeplabs collaboration with the National Technical University of Athens (NTUA).
• Student Projects
  • Maria Parelli's work on Interpretable Visual Question Answering Via Reasoning Supervision, presented at ICIP 2023 as a lecture.
  • Zacharias Anastasaki's diploma thesis on Self-Supervised Visual Relationship Detection, also partially presented at WAVC 2023 as a poster.
  • Danai Brilli's diploma thesis on Scene Graph Guided Visual Question Answering, presented at BMVC 2025 SU2HM Workshop both as an oral and poster presentation.

CVIA lectures

• Guess lecturer for the Computer Vision & Image Analysis (CVIA) Course.
• Part of the Interdisciplinary Space Master (ISM) / Master is Information and Computer Science (MICS) of SnT, Univesity of Luxembourg.

Teaching Assistant

• During my PhD, I served as a teaching assistant for multiple undergraduate modules of the School of Computer Science (University of Nottingham).
• My duties included lab tutoring, coursework marking and running tutorials.
• Modules
  • Computer Fundamentals
  • C++ Programming
  • Mobile Developments
  • Image Processing

• I was awarded a full funded PhD scholarship by the University of Nottingham and The Douglas Bomford Trust.

• I graduated with the 4th highest grade of my university cohord (DUTH).

CAD-Assistant

• This repository will contain the official codebase for our ICCV 2025 paper: CAD-Assistant: Tool-Augmented VLLMs as Generic CAD Task Solvers

CAD-Recode

• The implementation of CAD-Recode, a 3D CAD reverse engineering method introduced in our paper: CAD-Recode: Reverse Engineering CAD Code from Point Clouds

CPTSketchGraphs

• A dataset of 80 millon constraint preserving transformations (CPTs) of CAD sketches derived from the SketchGraphs dataset.

KeypointsToLandmarks

• Code for our TPAMI paper From Keypoints to Object Landmarks via Self-Training Correspondence: A novel approach to Unsupervised Landmark Discovery.

UnsupervisedLandmarks

• Code NeurIPS 2020 paper, Unsupervised Learning of Object Landmarks via Self-Training Correspondence.