Peter Bangert
Senior Platform Engineer · Berlin, Germany | Detroit, Michigan
I build and operate large-scale infrastructure platforms. My work spans Kubernetes,
bare-metal environments, low-latency networking, and systems programming in Rust.
Detroit sports fan
Experience
Senior Platform Engineer, Full-Time · Stelia Ltd
- Building a GPU PaaS/IaaS from the ground up
- Lead engineer for Kubernetes Operator development using a full Rust stack
-
Built a NUMA/CPU topology-aware GPU compute instance service
- Upstreamed a change to the Kubernetes Topology Manager to balance CPUs across sockets and pack them by L3 cache. Submit patches to cloud-init, virtiofsd, Kamaji
- Submitted presentations to Open Source Summit and Cloud Native AI Paris; presenting at the Leaseweb AI Summit
- Built a managed Kubernetes service using the Kamaji control plane and the GPU instance service
- Building managed inference with vLLM, the llm-d proxy, and RoCE RDMA
- Built NAT gateway, firewall, and L4 load-balancer services with IPVS Maglev hashing; exploring eBPF Unimog
- Managing the observability stack, GitHub Actions CI/CD pipeline, OS image build platform, and Flux GitOps
- Operating Rook Ceph for block, object, and shared filesystem storage
- Provisioning servers through BMC/PXE boot configuration and coordinating hardware delivery
Network Engineer · Scalable Capital
- Designed and operated secure network connectivity with financial partners, exchanges, and banks
- Delivered a six‑figure cost‑saving initiative centralizing organization‑wide ingress and egress traffic
- Improved security posture through unified traffic filtering and monitoring, supporting BaFin audits
Cloud Engineer · Scalable Capital
- Built and operated cloud infrastructure for a large‑scale mobile brokerage and wealth management platform
- Worked extensively with AWS and Terraform to manage production environments
Graduate Research Assistant · Technische Universität Berlin
- Conducted research in low‑latency networking and kernel bypass frameworks
- Worked on RDMA‑based congestion control and AI‑assisted optimization of MPLS stacks
Platform Engineer · Adjust
- Maintained Gentoo‑based infrastructure spanning 1300+ bare metal servers
- Developed and operated internal Perl‑based infrastructure‑as‑code tooling
- Worked with Kafka, Postgres, Nginx, Ceph, Aerospike, Flink, and DNS systems
Education
M.Sc. Computer Science
Thesis: “Reactive vs Proactive Distributed Stream Processing Autoscaling Solutions”Constructed a Flink/Spark cloud autoscaler titled TidalScale utilizing machine-learning time-series prediction for load forecasting. Built on Kubernetes.
B.Sc. Computer Science
Thesis: “Analysis of Classical and ML based Adaptive Video Streaming Algorithms”NS-3 based evaluation of deep-learning HTTP adaptive video streaming approaches.