Toronto, ON · York University

ERIC YU

Statistics Undergraduate & Data Analyst

Building statistical models, interactive web tools, and data-driven applications — the kind that turn a messy set of inputs into one number you can act on.

Sep 2025 → Apr 2029 · B.A. StatisticsNow · Faculty of Science Student Ambassador2 projects shipped & live
Projects

Two things I built end to end, both live.

A ski and snowboard gear matcher that turns body metrics, ability and conditions into a real spec sheet — then scores a catalogue against it.

You give it height, weight, ability, riding style, expected temperature and budget. It computes target ski or board length, waist width, boot flex, goggle lens VLT and jacket insulation, then ranks every item in the catalogue against that profile. Height sets the base length, riding style shifts it 3–10 cm, ability adds or removes up to 8 cm, and body mass fine-tunes by ±6 cm. Budget acts as a hard filter with exact-tier matches weighted heavily; conditions drive optics and insulation on their own axis. The point was to kill the zero-result search — you always leave with a ranked setup and the reasoning behind it.

Key features
  • Weighted probabilistic scoring over 50+ user and equipment parameters, normalised to a 0–100 confidence score with ranked runners-up.
  • A deterministic matcher — the same profile always returns the same spec sheet, which makes the model testable rather than magic.
  • An interactive SVG loadout visualiser: five body zones light up as helmet, goggles, jacket, boots and skis lock in.
  • Products are drawn, not photographed. Specs drive the render, so a 72 mm carver is visibly narrower than a 118 mm powder ski.
  • Retailer links are generated at request time across Evo, REI, Sport Chek, Backcountry and The House — nothing stored, nothing to rot.
  • Bilingual EN / ZH interface, built on 2025/26 season data.
At a glance
Catalogue
58 items
Brands
11
Categories
6
Budget tiers
3
Length model
Riding style
Ability
Base — height166 cm
Style — all-mountain0 cm
Ability — intermediate0 cm
Body mass+1 cm
Recommended
167cm
Waist
98mm

Simplified illustration of the sizing logic. The live matcher scores a full catalogue on top of this.

Stack
Next.js 14App RouterTypeScript (strict)Tailwind CSS 3.4SVGVercel

CryptoVault

cryptovault.site

A client-side-only multi-chain wallet: one recovery phrase, five independent key trees, and no backend to trust.

A self-custodial browser wallet spanning Bitcoin, Ethereum, BNB Smart Chain, Solana, XRP Ledger and TON. A single BIP-39 phrase derives five independent key trees, and every signature happens in the tab — there is no server, and no telemetry. Each chain implements one common adapter interface for balance queries, transaction building and signing, and the adapters are code-split so the initial bundle stays small no matter how many chains are supported.

Key features
  • Keys never leave the device. The vault is encrypted with AES-256-GCM and the password stretched with PBKDF2-SHA256.
  • Private keys are derived fresh for each operation and discarded the moment signing finishes.
  • Native sends including Bitcoin UTXO selection, plus ERC-20 tokens on EVM chains and SPL tokens on Solana.
  • One-click ERC-20 deployment on top of OpenZeppelin’s standard implementation.
  • Swaps routed through 0x on Ethereum and BSC, Jupiter on Solana, with a transparent 0.5% fee.
  • Every amount is held as a bigint in native units — sats, wei, lamports — so floating-point drift is impossible.
  • Tested against published derivation vectors, local EVM nodes, swap fee maths and Playwright end-to-end runs — none of which need a private key.
At a glance
Chains
6
Seed → key trees
1 → 5
Vault crypto
AES-256-GCM
Backend calls
0
Stack
Next.jsReactTailwind CSS v4Zustand@scure/bip39@scure/bip32@noble/curves@scure/btc-signerethers@solana/web3.jsxrpl@ton/tonPlaywright
About

I’m a second-year Statistics major at York University, where I also serve as a Faculty of Science Student Ambassador — which mostly means answering the questions I was asking a year ago.

Most of what I build starts as a statistics problem and ends as something you can click. Freezed began as a sizing formula and grew into a full matching engine; CryptoVault started as a question about key derivation and turned into a wallet that never phones home. I lean on AI co-pilots as a standing part of my workflow — for web development, architecture and scaffolding — which keeps my own hours on the parts that actually decide whether a project is any good: the model, the data, and the edge cases.

Off the clock I’m a freestyle skier — the reason Freezed exists at all — and I 3D print. Both are the same loop as the work: build to a spec, take it out, find where it fails, adjust the spec.

In short
  • B.A. Statistics, York University — second year
  • Faculty of Science Student Ambassador
  • AI co-pilots as a standing part of the workflow
  • Freestyle skier
  • 3D printing
Toolkit

Data & Analytics

Data ModelingProbabilistic ScoringAlgorithmic LogicStatistical AnalysisData Visualization

Languages & Databases

SQLPythonR (ggplot2, dplyr)TypeScriptGit

Frameworks & Tools

Next.jsTailwind CSSVercelExcel

Coursework

Statistical ModelingProbability DistributionsData Processing & HygieneApplied Mathematical Computing
Experience & Leadership
Jul 2026 → Sep 2026

Technical Specialist

Camp STEAM · Vancouver, BC

Ran daily technical operations and led technology programming for groups of 20+ youth, translating technical logic and data structures into something a twelve-year-old will actually sit through. Evaluated and troubleshot hardware and software to keep sessions running.

Sep 2026 → Apr 2027

Student Ambassador

York University, Faculty of Science

Selected as a departmental representative for campus showcase events — walking prospective students through program pathways, running Q&A sessions and tours.

Oct 2025 → Apr 2026

First-Year Representative

York University Ski & Snowboard Club

Organised trips and club events end to end, and read engagement and feedback data to schedule the next ones better.

Sep 2025 → Apr 2029

B.A. Statistics

York University · Toronto, ON

Statistical modeling, probability distributions, data processing and hygiene, applied mathematical computing.

Contact

LET’S BUILD SOMETHING.

Email is the fastest way to reach me — for data work, a project, or a question about anything on this page.