CV

Download the PDF version

Computer Science Ph.D. student at the University of Chicago researching efficient systems for caching, storage, memory management, and LLM serving.

Education

• Ph.D. in Computer Science, University of Chicago, Sep 2025 – Present
  • GPA: 4.00
  • Advisors: Prof. Haryadi S. Gunawi and Prof. Juncheng Yang (Harvard)
• B.S. in Computer Science, Bandung Institute of Technology, Jul 2021 – Jul 2025
  • GPA: 3.95

Publications

• Haocheng Xia, William Nixon, Bintang Dwi Marthen, Pranav Bhandari, Juncheng Yang. “Learning-Augmented Heuristics: Simple yet Smart, Robust and Interpretable Cache Eviction.” USENIX Symposium on Operating Systems Design and Implementation (OSDI), 2026.
• Ray A. O. Sinurat, William Nixon, Philip Carns, … Haryadi S. Gunawi, Hariharan Devarajan. “GLANCED-IO: Taming I/O Optimization for Deep Learning at Scale.” ACM International Symposium on High-Performance Parallel and Distributed Computing (HPDC), 2026.
• Ray A. O. Sinurat, William Nixon, Haryadi S. Gunawi, Nikoli Dryden, Hariharan Devarajan. “HORATIO: Bridging Management and Analysis of Traces at Scale.” International Conference on Scientific and Statistical Database Management (SSDBM), 2026.
• Samuel Cahyawijaya, Holy Lovenia, Joel Ruben Antony Moniz, … William Nixon, … Peerat Limkonchotiwat. “Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia.” Annual Meeting of the Association for Computational Linguistics (ACL), 2025.
• Holy Lovenia, Rahmad Mahendra, Salsabil Maulana Akbar, … William Nixon, … Samuel Cahyawijaya. “SEACrowd: A Multilingual Multimodal Data Hub and Benchmark Suite for Southeast Asian Languages.” Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024.

Preprints

• William Nixon, Muxin Tian, Yunjia Zheng, Haryadi S. Gunawi, Juncheng Yang. “From Requests to Sessions: A Large-Scale Characterization of Human-Driven Agentic Workloads.” In submission, 2026.
• William Nixon, Jon Durbin, Florian Standhartinger, Haryadi S. Gunawi, Juncheng Yang. “A Year in LLM Serving: Workload Evolution, Caching and Load-Balancing.” In submission, 2026. arXiv
• Hongshu Yan, Yunjia Zheng, William Nixon, Yazhuo Zhang, Pranav Bhandari, Haryadi S. Gunawi, Ana Klimovic, Juncheng Yang. “Demystifying and Optimizing Slab Rebalance in Memory Caches.” In submission, 2026.

Research experience

• Caching and Memory Management, Sep 2025 – Present
  • Learning-Augmented Heuristics, LAH (OSDI’26): uses ML to make simple cache heuristics adaptive by automatically tuning their parameters, outperforming more complex ML-based cache policies with negligible runtime overhead; improves cache efficiency by 26% over S3-FIFO and 8% over prior state of the art.
  • Adaptive Slab Rebalancing (preprint): improves memory allocation in slab-based caches by stabilizing noisy rebalance signals and preventing unnecessary slab movements; production deployment reduces DRAM misses by 20% and CPU utilization by up to 5%.
  • Implemented S4-FIFO and Adaptive-Marginal-Hits in C++/CacheLib and worked with Meta engineers to deploy and evaluate them in production.
• LLM Serving and Agentic Workloads, Jul 2026 – Present
  • Agentic Workload Analysis (preprint): analyzes 1.2M production requests, 205B input tokens, and 12K coding-agent sessions across 13 agent frameworks; characterizes tool execution, context growth, and KV-cache reuse, and releases the largest production agentic trace with tool- and block-level information to date.
  • A Year in LLM Serving (preprint): analyzes and releases a year-long production trace of 6.1B LLM requests across 9K+ models, the largest public LLM serving dataset to date, enabling large-scale studies of workload evolution, prefix caching, and request routing.
  • Led both projects: built the analysis pipeline for TBs of production requests and the event-driven simulator for cache eviction and request routing, and maintain the open-source trace and benchmark repositories.
• Deep Learning I/O and Trace Analysis, Mar 2025 – Jul 2026
  • GLANCED-IO (HPDC’26): optimizes I/O bottlenecks in large-scale weather-forecasting workloads by tuning I/O and data configurations through GPU-free proxy applications, improving training performance by up to 1.57x while avoiding expensive GPU runs during tuning. In collaboration with Lawrence Livermore National Laboratory (LLNL).
  • HORATIO (SSDBM’26): develops an out-of-place index for gzip-compressed JSONL traces emitted by scientific applications, enabling fast queries directly over compressed data without format conversion and with minimal additional storage; up to 83x faster analysis. In collaboration with LLNL.

Open-source contributions

• Google Summer of Code (GSoC), Open Source Research Experience (OSRE), UC Santa Cruz OSPO, Sep 2023 – Mar 2025
  • Built an extensible benchmark for concept drift in ML-for-systems workloads, comparing three online adaptation algorithms against retraining across four datasets with mentors from Argonne National Laboratory.
  • Integrated Matchmaker, DriftSurf, and AUE into ML-based caching and storage systems, including GL-Cache and LinnOS, reducing cache misses by up to 5%.
  • Released open-source implementations, experiment results, and a reproducible Chameleon Cloud artifact.
• SEACrowd, Contributor and Community Reviewer, Jul 2023 – Sep 2024
  • Implemented Hugging Face dataset loaders for under-resourced Southeast Asian languages as part of the SEACrowd open-data and benchmark ecosystem published at EMNLP’24.
  • Reviewed 50+ community contributions for loader correctness, dataset metadata, and interface compatibility.

Industry experience

• Backend Developer, Bukit Vista Hospitality Services, Mar 2022 – Dec 2022
  • Built the backend for an in-house multi-property availability calendar, contributing to API design, database modeling, implementation, and deployment on AWS.
  • Automated finance and HR workflows previously handled manually, reducing processing time by up to 90%.
  • Technologies: Node.js, Python, AWS.

Teaching

• CMSC 23000: Operating Systems, University of Chicago — Aut 2025, 2026
• CMSC 33200: Topics in Operating Systems, University of Chicago — Spr 2026
• CMSC 14400: Systems Programming II, University of Chicago — Win 2026

Skills

• Languages: C, C++, Python, Go, Java, JavaScript, SQL, Bash/Shell
• ML & parallel computing: PyTorch, TensorFlow, CUDA, MPI, OpenMP, SLURM
• Systems & data: CacheLib, Redis, Apache Spark, Pandas, PyArrow/Parquet, RabbitMQ, PostgreSQL
• Infrastructure & cloud: Linux, Git, Docker, Kubernetes, AWS, Google Cloud, Chameleon Cloud