2004
Love at First Infrastructure
The garbage truck that came by every week was my favourite public service (the only one I saw, so my dream job was to be a garbage collector). The obsession with cities starts early.
Data Analyst (in all but name) + Lifelong Transit Nerd = Business Intelligence Analyst (the best you'll ever hire)
I've been riding, advocating, analysing, and automating my way through cities since before I knew what a dataset was.
Every stop
shaped the route.
2005
Started Commuting!
Walked ten minutes to and from kindergarten daily.
2012
New Mode Unlocked: School Bus
Started taking the school bus to middle school. Learned that transit is social infrastructure where you chat with friends.
2016
An Iconic Grade 10 Class Presentation
We could present on anything we wanted. I chose to focus on reducing traffic congestion in Toronto.
- Argued for better transit as the primary solution to urban congestion — still holds
- Some prescient ideas (better tech, active transportation)
- Some ideas that didn't age so well (e.g. Metrolinx as a unified agency)
2017
MiWay Student Ambassador Program
Student Transit Ambassador at high school, promoting awareness of MiWay (public transit agency) and the Mississauga Moves transportation master plan. First real advocacy work.
2018
Parkette Proposal
Submitted plans to a city councillor to beautify an unused slip of electric utility land into a public parkette, connecting the neighbourhood to the bus stop on the arterial road. Got nowhere. Lesson: change is slow. But you still have to try.
2019
New Mode Unlocked: London Transit
Started university. Switched from car-dependent suburban Mississauga to car-dependent London, Ontario. But now I don't have my parents' car.
See more
- Started using Transit App because the arrival predictions were better.
- Bus service was half-decent around the university.
- Began mentally comparing transit systems across every city I visited
2020
First Taste of Serious Math
Received an Undergraduate Student Research Award. Learned about convexity and how math research can be applied to real-world problems.
Read the paper (PDF)2021
Machine Learning x London Transit Data
Built a full data pipeline to collect and clean real-time GTFS data from London Transit, then compared ML models for arrival time prediction. First hands-on experience with transit data formats and a useful lesson in what makes a model evaluation actually useful.
See more
- Ingested GTFS-realtime protobuf feeds on a cron schedule, handled timezone normalization and feed outages, and built a clean training dataset from scratch
- Compared regression, decision trees, and neural networks using R², which in hindsight doesn't capture what matters for scheduling (tail lateness, not average error)
- What I'd do differently now: Use MAE and percentile error metrics, treat severe delays as a separate classification problem, and define a clearer hypothesis
2021
Linear Hashing Talk — Math Club at Western
Gave a public talk to Western's Math Club breaking down a CS concept for a mathematical audience.
Watch the talk (Facebook)2022
Western Active Transportation Society
Advocated for walking, cycling, and sustainable mobility on campus and in the city.
2023
Published in Ivey Business Review
A strategy proposal to address Western University's housing woes.
Read the article2023
Project: Tech Layoffs Statistical Analysis
Applied beta regression to 2,400+ tech layoff records to model what drives the severity of workforce cuts — from exploratory data analysis through model selection to evaluation.
See more
- The analysis involved exploratory visualization, Cook's distance outlier removal, likelihood ratio tests for nested models, link function comparison (logit, probit, cauchit, etc.), and interaction term selection, evaluated with a train-validation-test split.
- What I'd do differently now: The project dropped ~33% of observations with missing response values without any sensitivity analysis, which probably introduces non-response bias. I'd now use multiple imputation or at minimum compare the dropped vs. retained populations. I'd also use zero-one inflated beta regression instead of naïvely capping boundary values, add temporal features rather than discarding the date column entirely, and avoid comparing AIC across models fit on different data subsets.
2023
CEOx1Day
Matched with Ashley Kirwan of Orix Geoscience! Spent a day in the C-suite learning how data informs real executive decisions.
Read about me in the news!2024
Singapore Rewired My Brain
Exchange semester in Singapore. Spent time biking, ridesharing, MRTing, and busing around the city state, returning to Toronto with a vision for what's possible.
2024
New Mode Unlocked: Bikeshare Toronto
Moved to downtown Toronto. Ditched the car entirely: became an avid bikeshare user, walker everywhere, and devoted follower of every transit project reshaping the city.
See more
- Bikeshare Toronto cumulative stats: 2000km and 500+ trips!
- Living through Line 5 (Eglinton Crosstown) and Line 6 (Finch West) openings!
- Staying curious about the King Street Transit Priority Corridor, Yonge Tomorrow, and Ookwemin Minising
2025
Full time job: Markets, Risk, Data, Automation
Completed various rotations on the trading floor and landed a permanent role in market risk. Data is at the heart of it all.
Highlight projects
- Automated repurchase agreement collateral allocator
- FX futures roll analysis and forecasting
- Swap pricing automation and daily interest rate news summary for senior salespeople
- Broker confirm email capture
- Annual limit review automation
2026
Now
Transit App: Business Intelligence Analyst
The next stop? Somewhere I can merge my passion for transit with my skills in all things data.
Things I've
actually built.
Some are serious. Some are delightfully hacky. All taught me something about turning a messy question into a working answer.
Mississauga Chess Club Website
Helped establish an online presence for the local chess club — website, Google Maps listing, contact flow. First time I built something for a real organization with real people depending on it.
MandarinLearn — Spaced Repetition App
A spaced repetition Mandarin learning app built for heritage speakers — people who grew up hearing the language but never formally learned to read or write it. Built with AI-assisted development throughout.
Details & link
- Targeted a specific underserved gap: heritage Mandarin speakers who need a different learning curve than total beginners
- Used AI tools throughout — content generation, curriculum logic, UI copy, iteration — while staying the one making product and architectural decisions
- Live and in use
Standard Notes Blog Frontend
Custom frontend for publishing notes from Standard Notes as a blog. Bridged a personal notes workflow with a public-facing reading experience.
Economics Exam Bank Tester
PDF OCR pipeline that ingests exam PDFs, semantically categorizes questions by type, randomizes them into practice sets, and serves them via a command-line interface.
Details
- OCR extraction from scanned exam PDFs; semantic categorization by question type, not just keyword matching
- Randomization engine for practice set generation
- Terminal UI — clean prompts, score tracking, hint mode
UWO Empty Room Finder
Scraped Western's class schedule system, stored availability in a database, and surfaced empty classrooms in real time via a CLI. Solved an annoying problem efficiently.
Details
- Scraped timetable data and room assignment schedules on a refresh cycle
- Designed a normalized SQLite schema to track room → time → availability
- CLI interface: query "empty rooms right now, building X, seats > 20"
Mustang Capital — Sports Betting Arbitrage Tracker
End-to-end accounting system for a group sports betting operation — tracked every bet, money flow, and bonus arbitrage across platforms, with P&L calculations and a running ledger.
Details
- Logged every bet placed across platforms with odds, stake, and outcome
- Tracked capital contributions per person and running balances
- Calculated real-time P&L and net settlement amounts between participants
- Identified available arbitrage and promotional +EV opportunities as they appeared
- Essentially a mini double-entry ledger — built intuition for financial data modeling
Market Participant Survey — Internal Report Prototype
Built an early internal report prototype for the Bank of Canada's Market Participant Survey — clean layout, summary tables, and visual hierarchy designed for senior stakeholders.
Power Futures Pricer
Power Trading Desk — pricing tool for power futures; inputs from the desk, pricing model applied, clean outputs ready for client conversations or trade decisions.
Automated STRIPs Offer Sheet Generator
Rates Trading Desk — daily automation that generates the STRIPs offer sheet each morning; pulls market data, applies the formatting template, and delivers a ready-to-distribute document without manual input.
Automated Rate Checks, PnL Forecasts & OSFI Review
Mortgage Investment Group — automated daily rate checks, PnL forecast generation, and OSFI review prep; removed manual steps from the team's critical morning workflow.
Commitment Data Visualizations & Business Explainers
Mortgage Investment Group — suite of visualizations covering market risk reporting, prepayments by maturity, and business explainer charts for internal stakeholders.
Bond Collateral Allocator
Repo Trading Desk — data pipeline that automates collateral allocation decisions based on configurable eligibility rules. Replaced a spreadsheet process that was, charitably, a liability.
Details
- Ingests daily position data, applies eligibility rules, outputs optimized allocation
- Configurable rule engine — business logic separated from code
- Generates historical allocation records for trend analysis
- Eliminated ~1 hour of daily manual reconciliation
FX Futures Roll Data Analysis & Backtesting
Quantitative Investment Strategies (QIS) Team — analysis and backtesting framework for FX futures roll strategies; pulled multi-source market data, modeled roll dynamics, and evaluated historical strategy performance.
Basis Swap Workflow Automation
Interest Rate Derivative Sales Desk — end-to-end automation covering RFQ, pricing, summary generation, and trade confirmation for basis swap trades; turned a multi-step manual process into a single-run workflow.
Broker Crossing Email Report
Options Trading Desk — pipeline that reads incoming broker crossing emails (each in a different format), parses trade details, stores everything in a database, and sends a daily digest. Turned an inbox into a data source.
Details
- Handled multiple broker email formats with different layouts — robust parsing, not brittle regex
- Backfilled historical emails on first run to build a complete trade record
- Daily report summarising all crosses confirmed that day, sent automatically each morning
- Replaced a manual process of reading and logging each confirmation by hand
Implied vs Historical Volatility Analysis
Options Trading Desk— analysis tool comparing implied and historical vol across positions; surfaces pricing dislocations and gives the desk a systematic lens on vol surface dynamics.
Client Trades Dashboard & Daily Email Report
Options Trading Desk — dashboard tracking client activity with a daily email summary; gave the desk an instant view of what clients were doing and what to follow up on.
Daily Market Risk Report — Generation & Data Upload
Market Risk — automated the full generation and upload pipeline for the daily risk report; end-to-end from data pull to formatted output and system upload, replacing a fragile manual process.
Webex Captions Capture
Market Risk — automation script that captures live Webex meeting captions and saves them as a structured transcript, while recording audio and using a second model to transcribe it for improved accuracy.
Details
- Hooks into Webex's accessibility caption feed and streams text to file
- Timestamps, speaker labels, and clean plain-text output
- Replaces a manual note-taking process for recurring meetings
Five lines of expertise.
One network.
Like any good transit system, the value is in how the lines connect.
Workflow Automation
Python, SQL, VBA, etc.
Web Scraping
OCR Application
CLI Tools
Workflow Automation
CLI Tools
Statistics and Machine Learning
Probability and Statistics
Hypothesis Testing
Exploratory Data Analysis
Predictive Modeling
Study Design
Data to Business Translation
Dashboard Design
Stakeholder Reporting
Data Storytelling
Problem Translation
Data Quality Standards
AI Tools
LLM Workflow Integration
Prompt Engineering
Semantic Embeddings
AI-Assisted Coding
Transit & Urban Domain
Active Transportation
Rider Experience
GTFS & Ridership Data
Urban Planning
Transportation Advocacy
Let's talk about transit data.
Or urban planning for better cities. Or how we can improve bikeshare.