Software Engineer & AI/ML Researcher working across production software, graph machine learning, temporal graphs, trustworthy ML, and high-performance GPU systems.
I currently work as a Software Engineer at WeDoSoftware and will begin the MSc Advanced Computing at Imperial College London in September 2026. My research has resulted in first-author work at KSEM, ICSC, IcETRAN, INFOTEH-JAHORINA, and IJEEC. My TempoTriads research received the 2026 Annual Award in Computing from the Mathematical Institute of the Serbian Academy of Sciences and Arts.
| Project | What it does | Highlights |
|---|---|---|
| FuseSampleAgg | Fused PyTorch CUDA operator for neighbourhood sampling and mean aggregation | 2.24–3.48× lower step latency and up to 160× lower transient GPU memory; KSEM 2026 |
| TempoTriads | Streaming estimator for temporal triadic motifs | Bottom-k sampling, Horvitz–Thompson weighting, dependence-aware confidence intervals; ICSC 2025; national award |
| AutoSAGE | Input-aware scheduling for sparse CUDA GNN kernels | Guarded runtime selection for SpMM, SDDMM, and CSR attention; up to 4.7× kernel-level speedup |
| Fuse-T | Gated late fusion of RoBERTa text semantics and reply-graph topology | Unseen-event rumour classification; IcETRAN 2026 |
| ECHO-Proteins | Reproducible edge-aware GNN baselines on OGBN-Proteins | Calibration, threshold selection, and population-held-out evaluation; IJEEC 2026 |
- Software Engineer, WeDoSoftware — multi-tenant SaaS, distributed background processing, identity integrations, observability, production releases, technical interviews, and intern mentoring.
- Research Engineering Intern & External Collaborator, Fudan University DataNet Lab — C++/Python graph analytics, performance engineering, testing, packaging, and dataset infrastructure.
- Undergraduate Teaching Assistant, University of Novi Sad — Parallel Programming with Intel oneTBB and OpenCL; Object-Oriented Programming II in C++.
Languages: Python, C++, C#, Java, TypeScript, SQL
ML & systems: PyTorch, PyTorch Geometric, CUDA, NumPy, graph neural networks
Software engineering: .NET / ASP.NET Core, Angular, Spring Boot, PostgreSQL, Docker, Azure
Platform & reliability: Kubernetes, OpenShift, OpenTelemetry, Elasticsearch, Kibana, GitHub Actions
Graph machine learning · ML systems · temporal graphs · graph analytics · trustworthy AI · high-performance computing