Experience
Research Assistant
- Built a Python data ingestion SDK handling multiple schema types with alias-based column resolution, automated type coercion, and input sanitisation.
- Resolved out-of-memory crashes by migrating data pipelines from pandas to Polars LazyFrames, enabling national-scale dataset processing.
- Achieved 99.9997% algorithmic fidelity against legacy implementations and improved Roth-Peranson matching performance by 30x on 300k+ candidates via heap-based priority queues.
- Wrote 1,500+ lines of validation tests including legacy parity checks, edge case coverage, and preference ordering verification.
- Collaborated with PhD researchers to write, debug, and implement custom software solutions for spatial data analysis.
Full-Stack Developer
- Designed, developed, and maintained a highly performant, responsive website powered by Astro, Preact and Tailwind.
- Integrated custom open-source CMS (Kea) to provide a visual editor for non-technical staff to manage content.
- Led the full SDLC, partnering with non-technical stakeholders to iteratively define and ship features.
Software Engineer
- Developed custom ETL automation pipelines to process ticket sales, integrating Humanitix API with Firebase and Google Sheets, eliminating 100% of manual data entry operations.
- Built an interactive, crowdsourced music-voting platform using SvelteKit and Firebase to maximise attendee engagement prior to the event.
- Designed, developed, and maintained the official event website, driving ticket sales and raising NZ$6,000 for UNICEF.
Projects
PDF Editor (Browser-Native WebAssembly Engine)
- Built a client-side PDF editor in Svelte 5 and TypeScript, integrating a Rust WebAssembly runtime for Google PDFium with an LRU page cache and unproxied binary stores to prevent memory exhaustion on large documents.
- Engineered a 4-pass spatial clustering algorithm to reconstruct semantic paragraphs from raw PDF glyphs, enabling in-place text replacement and automatic line reflow while preserving embedded font subsets.
- Implemented the W3C WebMCP draft specification for autonomous AI agents, exposing 47 atomic tools over a Semantic Document Graph with single-writer mutex locking, byte-level anchor verification, and staged proposals.
- Architected an extensible canvas element registry and serialisation pipeline baking vector paths, rich text, and AcroForms directly into native PDFium content streams as standard PDF objects.
- Designed an invertible command architecture supporting transactional undo and redo stacks, integrating 1,000ms sliding-window merge coalescing for continuous interactions and compound rollback on failed mutations.
- Implemented an ISO 32000 compliant digital signing pipeline with client-side SHA-256 byte-range hashing and detached PKCS7 signatures, coupled with an automated sanitisation engine purging metadata, scripts, and hidden object streams.
- Built a client-side layout imposition engine supporting saddle-stitch booklet folding with mathematical folio pairing, N-up multi-page grid layouts, and multi-sheet poster tiling via pdf-lib.
- Developed a tabbed multi-document workspace featuring drag-and-drop page reorganisation, arbitrary page splitting, and a modular exporter generating flattened PDFs, high-DPI raster bundles, and plain text.
Utility Outages (Unified Monitoring Platform)
- Engineered the infrastructure, backend and scraper for a comprehensive, real-time utility outages monitoring platform, aggregating real-time power outages from 28+ providers, NZTA road events and more into a unified geospatial database.
- Architected a high-performance async REST API using Robyn and SQLAlchemy to serve filtered GeoJSON data.
- Developed a Python data scraper using Polars and Shapely to fetch, process, and normalise complex geometric data.
- Designed and integrated a geospatial time-series database architecture using PostgreSQL, TimescaleDB, and PostGIS to efficiently query points, polygons, and multilines.
- Containerised and orchestrated backend services, databases, and the data pipeline using Turborepo, Docker and Docker Compose in a hybrid monorepo deployment architecture.
Polymarket Trading Bot (Automated HFT Engine)
- Engineered an automated trading engine in Python and Rust for Polymarket crypto binary prediction markets executing breakout, copy trading, and flash crash strategies.
- Developed a native Rust VWAP engine tracking Spot Fiat feeds across Tier-1 exchange WebSockets (Coinbase, Kraken, Bitfinex), gaining a 5.6-second latency advantage over Chainlink oracle updates.
- Designed an upstream lock-based budget gating engine with half-compounding profit allocation, automated drawdown circuit breakers, and on-chain CTF redemption via gasless EIP-712 transactions.
- Built a high-frequency tick recorder and vectorized backtesting pipeline storing market orderbooks in chunked Parquet datasets for low-latency simulation.
Ucord (Model Context Protocol AI Client)
- Built an autonomous agentic client in TypeScript and Bun, implementing the Model Context Protocol (MCP) to standardize tool-calling capabilities across LLMs.
- Engineered a custom regex-based parser leveraging XML and JSON hybrid syntax, enabling structured tool execution for open-weight NVIDIA NIM models lacking native function-calling support.
- Architected a dynamic context manager with an image-description caching layer, seamlessly integrating LLaMA Vision models to process multimodal inputs within strict token budgets.
- Designed a stateful generation controller utilizing
AbortControllerto handle asynchronous chat interruptions, allowing the agent to gracefully pause, queue, and recalculate interactions.
Skills
Languages
- Python
- TypeScript
- JavaScript
- SQL
- Rust
- Java
- HTML/CSS
Frontend
- React
- Next.js
- Svelte/SvelteKit
- Tailwind CSS
Databases
- PostgreSQL
- PostGIS
- TimescaleDB
- Redis
- Firebase
- Cloudflare D1
Infra & DevOps
- Docker
- Docker Compose
- Cloudflare Workers
- Git
- GitHub
- CI/CD (GitHub Actions)
Data & Tools
- Polars
- Pandas
- Parquet
- NumPy
- Playwright (Automated Testing)
- Postman
- REST APIs
Education
University of Canterbury
Bachelor of Science (Computer Science & Statistics)
Jan 2025 - Expected Dec 2027 | Christchurch, NZ
Leadership & Community Involvement
Marketing Manager
- Led promotional strategies and designed marketing assets for university-wide tech workshops and industry networking, increasing student-sponsor engagement and event turnout.
Youth Leader
- Represented refugee youth at national forums (NRRF), advocating for inclusive policy development and improved access to education and social services, and facilitated leadership workshops and social activities for regional refugee communities.
TKAM Tuakana Student Mentor
- Provided peer mentorship to junior University students, facilitating workshops and community engagement activities to support university transition.
Youth Advisor
- Co-developed and piloted a national youth mental health programme, providing advocacy that influenced major organisational project decisions.
UC Global Leaders Programme
- Selected for an exclusive cohort focused on cross-cultural leadership, global civic engagement, and international collaboration.
Academic Awards:
SaaSathon 3rd Place Winner
Deloitte Hackathon Winner (Prioritising Tech for People and the Planet)
NZQA Outstanding Scholarship in Technology (Top 0.3% nationally)
Scholarships:
Te Kakau a Māui Scholarship (1 of 30 full-tuition)
Outstanding Impact Scholarship (1 of 70)