Curriculum vitae
Software Engineer · Applied AI & Research
Summary
Software engineer with eleven years in production and numerical software, now working on AI evaluation and computational research. Backend systems for European institutions since 2018; PhD candidate at the Romanian Academy in fractal and topological methods for medical image analysis; independent practice since February 2026. I care about whether a system’s measurements can be trusted — and I report the results that say no.
Core strengths
AI evaluation & experimental design
Matched-overlap stress tests, preregistered decision gates, paired statistics, and reporting the result even when it is negative.
Agentic systems & verification
Bounded agent missions with independent evaluators, seeded contradictions, critical vetoes and recorded trajectories.
Production backend & data engineering
European institutional systems since 2018 (Java, Spring, Oracle/SQL); Python services with FastAPI, SQLAlchemy, PostgreSQL/PostGIS and Docker.
Numerical & scientific computing
C/C++ and Fortran engineering software, analytic derivatives checked against finite differences, fractal and topological image analysis.
Scientific communication
Interactive explainers and this site: exact, tested models presented so a reader can check the claim themselves.
Experience
Independent practice
Software & AI consultant
Confidential client engagements and own research in AI evaluation, agent monitoring and Python services (FastAPI, async SQLAlchemy, Pydantic, PostgreSQL, Docker, pytest). Builds evaluation pipelines for coding agents that compare deterministic checks, LLM reviewers and escalation on held-out cases, measuring detection, false alarms, latency and cost. See Consulting for how engagements run.
European Commission · DG RTD
External ICT Contractor
Backend and database work on SEP Evaluation, the system independent experts use to review grant proposals and submit evaluation reports and scores. Java, Spring and Oracle/SQL on WebLogic: data-model changes, validation rules, permissions, dynamic forms and evaluation-workflow state transitions; structured application forms for Horizon Europe calls; live monitoring, incident resolution and controlled releases during deadline-critical calls.
European External Action Service
Software Engineer
Designed and developed a Java-based Security Portal with access control and data protection, adapting it to changing security and operational rules from international stakeholders. Brussels.
Tech-Code
Software Engineer
Java web applications and front-end components (Java, SQL, HTML, CSS, JavaScript), with attention to integration, maintainability, testing and technical documentation. Iași.
Bentley Systems
Software Developer and Structural Engineer
RM Bridge engineering software: C/C++ application components and a legacy Fortran calculation engine. Localised numerical defects without a usable debugger by recomputing expected results by hand, reproduced client-reported failures with engineering users, and translated updated U.S. bridge-design standards into production calculation rules validated against independent calculations. Iași.
Education
- PhD candidate, Informatics — Romanian Academy (SCOSAAR)
- MSc Civil Engineering (Structural Engineering) — Gheorghe Asachi Technical University of Iași
- BSc Civil Engineering — Gheorghe Asachi Technical University of Iași
Publications
- Lucian Alexandru Murgu, Tudor Barbu. Digital Topology Instability under Exact Decimation of Retinal Vessel Annotations: Stage Localization and Source Provenance. IEEE E-Health and Bioengineering Conference (EHB 2026), 2026.
- Tudor Barbu, Lucian Murgu. Multi-Task U-Net Architecture for Tumor Segmentation and Recognition based on Fractal Analysis and Nonlinear Reaction-Diffusion. Proceedings of the 21st Conference on Computer Science and Intelligence Systems (FedCSIS), Annals of Computer Science and Information Systems 47, 2026, pp. 191–196. doi:10.15439/2026F4871
- Lucian Alexandru Murgu, Tudor Barbu. PDE-Refined Local Fractal Dimension Prior Conditioning and Topology-Aware Refinement for Retinal Vessel Segmentation with a Swin-UNet-Style Backbone. Applied Sciences, 2026, 16(11), 5559. doi:10.3390/app16115559
- Lucian Murgu, Tudor Barbu. Deep Learning-Based Framework for Retinal Vessel Extraction Using Fractal Analysis and Nonlinear Diffusion. Advances in Digital Health and Medical Bioengineering II, IFMBE Proceedings (Springer), 2026, pp. 204–212. doi:10.1007/978-3-032-23952-5_22
- Tudor Barbu, Lucian Murgu. Comparison Between Novel Machine and Deep Learning-based Skin Tumor Recognition Techniques. 2024 E-Health and Bioengineering Conference (EHB), IEEE, 2024, pp. 1–4. doi:10.1109/EHB64556.2024.10805703