cv

Education

Ph.D. in Computer Science and Engineering

University of Nevada, Reno
Expected Dec 2026

M.S. in Computer Science and Engineering

University of Nevada, Reno
Expected Dec 2026

B.Sc. (Hons) in Computing Science, First Class Honours

Ulster University, Belfast, Northern Ireland
May 2017

Research Experience

Graduate Research Assistant

Intelligent Data and Systems Lab, University of Nevada, Reno
07/2021 – Present
  • OPSERVE: Built a malleable LLM serving system that combines persistent and transient GPU workers in batch-scheduled HPC environments, dynamically adapting prefill/decode capacity and preserving completed prefill state through shared KV storage. In evaluated workloads, OPSERVE reduced p95 end-to-end latency by 1.5–2.1× under sustainable load and by up to 14.3× under overload. Accepted to AIonHPC @ SC26, 2026.
  • BatchFlow: Built a distributed data-pipeline service for multi-job training that dynamically allocates shared data workers and cache capacity using online profiling and benefit-aware scheduling. A PyTorch-compatible prototype evaluated on AWS improved aggregate training throughput by up to 3.8× and cost efficiency by up to 2.1×. Accepted to SC26, 2026.
  • zkInfer: Designed a distributed proof-serving architecture for verifiable ML that decomposes inference into independently executable zero-knowledge proving jobs with resource-aware scheduling, memory-aware admission, and reusable proving artifacts. In evaluated workloads, decomposition reduced latency by up to 5.6× on one worker and distributed execution by up to 69×, while reducing peak per-machine memory by up to 48×.
  • HDRFS: Contributed machine-learning and data-pipeline work for multimodal wildland-fire sensor data as part of the NSF EPSCoR Harnessing the Data Revolution for Fire Science project, and presented research at annual project meetings over the past three years.

Ph.D. Research Intern

Nokia Bell Labs, Stuttgart, Germany
08/2022 – 11/2022
  • Designed and implemented a distributed zero-knowledge proof-generation system using Apache Spark.
  • Parallelized Number Theoretic Transform and Multi-Scalar Multiplication workloads across cluster resources to investigate proof-generation scalability and latency.
  • Characterized compute, memory, and parallelization bottlenecks in large-scale cryptographic workloads and presented findings to Bell Labs research teams.

Publications

Patrick Watters, Xiaolong Ma, Syed Zawad, Jingyuan Zhang, Yue Cheng, Lei Yang, Feng Yan. “BatchFlow: Benefit-Aware Data Pipeline Allocation and Batch Reuse for Multi-Job Training.” SC26, 2026. Accepted, to appear.
Patrick Watters, Xiaolong Ma, Harry Yu, Ian Foster, Michael Papka, Lei Yang, Feng Yan, Rajkumar Kettimuthu. “OPSERVE: Opportunistic LLM Inference over Fragmented GPU Capacity in HPC Systems.” AIonHPC @ SC26, 2026. Accepted, to appear.

Industry Experience

AI Model Evaluator

Outlier
2026 – Present
  • Evaluate conversational and voice AI systems through structured comparative assessment, focusing on response quality, interaction quality, task completion, and model behavior.

Senior Professional Services Consultant

Automated Intelligence, Belfast, Northern Ireland
07/2016 – 08/2021
  • Led enterprise and public-sector cloud migration projects involving Microsoft Azure and Microsoft 365, modernizing legacy data systems and workflows.
  • Designed and implemented C#/.NET and Azure-based systems for large-scale data migration, transformation, and automation.
  • Delivered production systems from requirements and architecture through implementation and deployment.

Teaching and Mentoring

Undergraduate Research Mentor

Research Experiences for Undergraduates, University of Nevada, Reno
06/2023 – 09/2023
  • Mentored undergraduate researchers developing weakly supervised object-detection models for wildfire smoke detection.
  • Guided implementation, GPU training, experimental analysis, and interpretation of research literature.

Graduate Teaching Assistant

Abstract Data Structures, University of Nevada, Reno
07/2021 – 07/2022
  • Taught C/C++ programming, data structures, algorithms, memory management, and complexity analysis.
  • Mentored students through office hours, debugging, and code review.

Technical Skills

Languages
Python C++ Rust C#
ML / Systems
PyTorch vLLM TensorFlow ONNX gRPC Apache Spark
Infrastructure
AWS Microsoft Azure Docker Linux HPC clusters

Awards and Honours

  • Automated Intelligence Values Award, 2020
  • Ulster EDGE Award, 2017
  • Dean's List, Faculty of Computing, Engineering and the Built Environment, Ulster University, 2017