Tolga Bilgis

HPC Intern, AI Infrastructure

• Automated containerized Linux/HPC infrastructure for the CIRCE AI platform with Ansible, GitLab CI/CD, Slurm, and Podman.
• Benchmarked five vector database backends using 2.7 million 4,096-dimensional vectors, evaluating throughput, latency, scalability, and resource use.
• Identified a retrieval backend delivering roughly 20× the existing setup’s throughput in the evaluated workload for an HPC-hosted RAG platform using a 2 TB arXiv corpus.

HPC System Administrator

• Administer the Rocky Linux student HPC cluster; configure distributed workloads and tune performance using Slurm, MPI, HPL, and HPL-MxP.
• Co-built and performance-tuned a heterogeneous cluster with NVIDIA A100 GPUs. Served as system administrator at SCC25; returning for SCC26.
• Co-built a 16-node Raspberry Pi cluster for student cluster competition work.

AI Coding / Math Expert

• Wrote, tested, reviewed, and refined programming and mathematical solutions in C++, Python, Java, Kotlin, and JavaScript.

Undergraduate Robotics Researcher

• Integrated sensors using Python and C++ and built real-time telemetry displaying a robotic vehicle’s speed, position, and sensor readings for a digital twin.

• Built a three-node Ubuntu Kubernetes cluster, automated host configuration with Ansible, and added GitHub Actions validation for Ansible.
• Deployed Prometheus, Grafana, and local persistent storage; tested pod replacement and a worker failure that demonstrated node-local storage limits.

• Built a TCP client/server tool measuring round-trip latency and sending throughput, with configurable host, port, duration, and buffer size.
• Implemented monotonic timing, min/average/max and p95/p99 reporting, and socket I/O handling partial transfers and interrupted calls.

• Split detection, facial recognition, and web alerts across two NVIDIA Jetson devices with local inference; earned 2nd Place, Best Use of Hardware at CUHackit 2026.