CV

Download PDF

Hamilton, New Zealand ·ezra_palmers@hotmail.com ·github.com/EzraPalmers

Profile

Honours student in Artificial Intelligence at the University of Waikato, and top-performing graduate of the undergraduate data analytics programme. I came into machine learning through statistics and probability theory - proof-level probability and stochastic processes, Bayesian inference, and applied statistics in R - and built deep learning, streaming ML, and GPU performance work on top of that foundation. My approach is to measure before optimising, attribute every gain to a specific cause, and report negative results honestly. Currently working on an NVIDIA-mentored GPU research project and an honours dissertation in EEG-based flow-state detection; a Master's or PhD in 2027 is a live option.

Education

BSc (Hons) in Artificial Intelligence
University of Waikato, 2026 (in progress)
Semester A 2026: Extremely Parallel Programming (99.05), Graph Theory (99.28), Deep Learning (99.14), ML for Data Streams (92.74).
Semester B 2026 (in progress): NVIDIA GPU research project (COMPX577), Interpretable Machine Learning (COMPX521), honours dissertation (AIMLX591, full year).
BSc, Major: Data Analytics
University of Waikato, conferred 15 April 2026
All nine 2025 papers graded A+ (average approximately 97%); 2024B foundation papers Machine Learning (94.93) and Principles of Probability and Statistics (96.10).
The University of Waikato issues letter grades only - there is no numeric GPA. A+ at Waikato means 90% or above; every 2024-2026 paper listed here is A+.

Awards

Current research

GPU research project with NVIDIA mentor
COMPX577, 2026 B, in progress. Honours research project mentored by an NVIDIA senior AI DevTech engineer, working on directions from NVIDIA's CUDA Core Compute Libraries team - production GPU-primitive design, continuing on from earlier CUDA kernel work. Early stage; this entry will grow as the project develops.
Honours dissertation - EEG flow-state detection
AIMLX591, 2026, 30 points, full year, in progress. Continues the funded summer research project below with the same supervisors, with an emphasis on pipeline validation and rigorous evaluation rather than headline accuracy.

Projects

Full descriptions are on the projects page. In brief: GPU/parallel performance engineering (CUDA, OpenCL, Hadoop, Spark), EEG flow-state research, deep learning on real-world image and audio data, a custom Radial Basis Function Network benchmark, streaming ML, a spectral graph theory report, and a local ASR/agent-tooling project (whisper-transcriber).

Work experience

Tutor, University of Waikato
DATAX222 Principles of Probability and Statistics (2026 B) - weekly 1-hour teaching tutorial, working probability problems on the whiteboard and fielding questions, in the paper I took at A+ in 2024.
DATAX201 Practical Data Science (2026 A) - two weekly 2-hour computer workshops, R fundamentals, 1-on-1 support and weekly quizzes/worksheets.

Skills

Statistics and probability
Proof-level probability and stochastic processes (Markov chains, PGFs, branching processes, MLE); Bayesian inference (INLA, regularisation as priors); applied statistics in R (GLMs, ARIMA, resampling); teaching probability at tutorial level.
GPU, parallel, and low-latency programming
CUDA kernel engineering and micro-benchmarking; OpenCL pipeline optimisation; Java parallelism (ForkJoinPool, CompletableFuture); a measure-hypothesise-test methodology and hot-path memory discipline. Hadoop/Spark experience is real but framed as data engineering, a different domain. C++ is a developing skill - some CUDA C++ and tutorials, no full standalone programs yet; strengthening through the current NVIDIA project.
Python and machine learning
End-to-end PyTorch pipelines (transfer learning, LoRA, mixed precision); from-scratch CNN work; systematic hyperparameter search with unbiased evaluation splits; streaming ML (drift detection, online ensembles); research data-pipeline engineering (signal processing, reproducible pipelines); custom scikit-learn-style estimators.
Languages
Python is the primary language and the strongest by a clear margin - it has always been the language of my ML work. Java: solid working knowledge, not a language I choose to build software in. C++: basic, growing through CUDA work.
AI and agent tooling
Build custom Claude Code skills and slash commands to automate multi-step workflows, and use disciplined project scaffolding (CLAUDE.md, handoff notes, prompt conventions) to keep agent-assisted work organised. Daily user of both Claude Code and Codex. Local ASR (faster-whisper) and local-LLM pipeline engineering with honestly-reported capability findings.
Other
Data storytelling: Tableau dashboards, presentations to non-expert audiences, Manim-animated technical talks, a conference poster. Web fundamentals: HTML/CSS/JS, PHP, MySQL, ER modelling. No cloud platform experience; Docker is understood but only lightly used personally, on a home Linux server. Outside of study: I run a small homelab (self-hosted services on a Raspberry Pi) as a personal interest.

Referees

Available on request.