I’m a software infrastructure engineer based in Dresden, Germany. My work sits across distributed systems, HPC scheduling, and the software that configures and controls real machines.
I came to software through physics, then studied software engineering and computer science in Tunisia and Germany. Algorithms and systems became the thread connecting them: understand the model, inspect the constraints, then build something that behaves predictably.
At Sparxity (formerly SpiNNcloud), I worked on runtime infrastructure, topology-aware allocation, Slurm integration, and machine discovery for SpiNNaker2. At Rohde & Schwarz, I investigated performance and timing problems in signal-processing libraries, Linux guests, and networking systems.
In my own projects, I make these ideas easier to inspect. My public work includes the topology-aware workload-management project and SchedLab, an interactive scheduling laboratory. This site is where I document their architecture and the decisions I’m still working through.
Contributed to the runtime and infrastructure of a SpiNNaker2 neuromorphic supercomputer: approximately five million cores, 720 interconnected boards, and a 2D hexagonal torus.
Drove architecture and implementation across three hardware allocator variants, coordinating scheduling, topology, resource state, and physical machine configuration.
Built deterministic placement heuristics using sliding windows, recursive free-region partitioning, DFS traversal, wrap-around coordinates, and fragmentation-aware scoring.
Proposed and drove the move toward Slurm integration while preserving native scheduling authority and separating board placement into a prolog-stage engine.
Developed C/C++ discovery, reachability, routing, UDP host-to-device communication, and diagnostic logging; worked on transactional allocation, release, and reconciliation.
Contributed to physical machine partitioning: board-to-board connectivity, link enablement, and isolating regions before discovery.
Worked through YAML-backed inventory with POSIX locks toward database-backed state, synchronized queues, and transactional allocation mappings.
Collaborated with runtime and infrastructure teams and researchers at Sandia National Laboratories and the University of Manchester.
Jul 2024 — Mar 2025
Rohde & Schwarz
Dresden · Germany
Working Student · Software Engineering
Performance engineering, Linux debugging, and test automation for networking and signal-processing systems.
Improved algorithms and data structures in a company-wide signal-transformation library, achieving 2.5× faster execution.
Investigated QEMU guest applications and PTP synchronization failures in O-RAN; traced timing shifts and mixed-traffic interference.
Benchmarked DPDK sender warm-up and jitter, and built Python, Pytest, and Playwright workflows spanning unit, integration, and end-to-end checks.
Mar — Jun 2022
Independent · Upwork
Remote
Freelance C++ / JavaScript Developer
Designed and implemented more than 50 interactive levels for an educational game teaching C++.
02 / Technical practice
The tools, and the thinking.
01 /
Systems & infrastructure
The software between a workload and the machine: resource state, allocation, concurrency, and hardware bring-up.
C / C++PythonLinuxDistributed systems
02 /
Scheduling & placement
Slurm integration, topology-aware allocation, graph algorithms, and the tradeoffs between locality and fragmentation.
SlurmSPANKGRES / LicensesGraph algorithms
03 /
Backend & reliability
Clear API contracts, transactional inventory, repeatable tests, and the diagnostics needed to understand a failure.
FastAPIPostgreSQLConcurrencyCI / CD
Other tools I’ve used include Docker, QEMU, DPDK, Wireshark, systemd, journalctl, Bash, Pytest, Playwright, and GitHub Actions.
03 / Education
A foundation in physics & software.
2024 — 2025 · Germany
Exchange · Computer Science
Ilmenau University of Technology
2021 — 2024 · Tunisia
MSc · Software Engineering
Tek-Up University
2017 — 2020 · Tunisia
Bachelor · Physics
Faculty of Sciences of Tunis
Algorithms, off the clock.
Three years of competitive programming and more than 600 problems solved. At IEEEXtreme 17.0 in 2023, my team ranked 21st of 1,048 in Tunisia and 849th of 8,263 worldwide.