Working Student (m/f/d) – High-Throughput Data Ingest / Networking (8 h/week)

We, the Chair of Biomedical Physics, develop a data ingest pipeline for CT detector readout with target data rates exceeding 300 Gbit/s. Data arrives as packetized UDP streams from FPGA-based hardware. Initial focus is multi-lane scaling using standard Linux networking (multi-queue NICs, RSS, parallel receive). Long-term goal is evaluation and prototyping of kernel-bypass approaches for full-rate operation. Responsibilities: • Design and benchmark a multi-lane UDP ingest pipeline • CPU/IRQ pinning and NUMA-aware memory allocation • Analyze bottlenecks using Linux performance tools (perf, ethtool, /proc, etc.) • Evaluate and prototype kernel-bypass alternatives (e.g. AF_XDP, DPDK) for extreme data rates • Document performance limits and scaling behavior Required Skills: • Solid C or C++ (systems-level programming) • Good understanding of Linux, networking (Ethernet/IP/UDP) and multithreading • Interest in performance engineering and low-level systems Nice to Have: • Experience with high-speed networking (RSS, NIC queues, zero-copy) • Familiarity with AF_XDP, DPDK, or RDMA • Understanding of NUMA and PCIe topology • Experience with high-throughput storage (NVMe, io_uring, SPDK) What You’ll Gain: • Hands-on experience with 100G+ networking • Performance engineering at hardware limits • Possibility for a thesis How to apply: Please send your CV and transcript of records to: benjaminraphael.berger@tum.de or lisa.marie.petzold@tum.de