About
Hi, I'm Shiyan, a Machine Learning Software Engineer based in Canada. I earned my degree in Computer Engineering from Queen's University, where I specialized in the Innovation & Entrepreneurship Stream.
I have experience in both machine learning and full-stack development. Currently, I’m developing LLM-based marketing products at TELUS' AI Accelerator as an ML Engineer. Before that, I worked at Shopify as a Backend Engineer, building APIs and checkout extensibility prototypes. I also spent time at Wealthsimple as a Full Stack Engineer, creating features to help users trade and invest in crypto.
Outside of work, I run Studio 50-50, a design and development shop that builds websites for small businesses and startups. In my free time, you can find me taking photos, playing music, cycling, and reading a good book.
Here are some technologies I've been working with recently:
- Python
- Java
- MySQL
- NextJS
- Typescript
- React
Where I've Worked
Machine Learning Engineer @ TELUS AI Accelerator
- Led the development of a Personalized Marketing Campaign Generator using RAG LLM architecture with Python, ACS Vector DB, Langchain, LiteLLM, GCP, Streamlit, and OpenAI GPT, reducing campaign creation time by 40%. Collaborated with stakeholders to define objectives, AB tested features and implemented a project roadmap.
- Developed an alerting and monitoring system for department-wide GenAI use cases, enhancing system reliability and performance during BFCM using Big Query, SQL, GCP, and Pulumi, resulting in a 12% reduction in downtime and faster issue resolution by 30%.
- Developed foundational features for the New API Marketplace using React, TypeScript, and NextJS, improving experience and performance.
- Optimized API requests by implementing advanced caching mechanisms with NextJS, significantly improving response times.
- Led a team of 10 engineers as National Lead for the co-op technical mentorship program, supporting +100 participants year-round.
Software Engineer Intern @ Wealthsimple
- Increased crypto engagement by 13% by developing crypto autobuy feature in React Native, Typescript, and Ruby on Rails as part of the Crypto Engineering team.
- Avoided +$10,000s of overage charges by developing CryptoCompare API request limit alerts with Ruby on Rails and Datadog.
- Saved 2 hours/week by independently developing the first automated end-to-end tests for Crypto mobile in Typescript and Appium.
Software Engineer Intern @ Shopify
- Developed and ran 20 load tests to simulate Checkout Classic scenarios, and analyzed results using Splunk and Datadog.
- Improved BFCM performance by 17% by detecting and optimizing N+1 GraphQL queries by batching, memoizing, and refactoring.
- Contributed to AccessScopes and Metafields prototypes by developing backend logic to gate writing to metafields with Ruby on Rails.
- Refactored prototypes based on feedback and shipped changes into production for the Checkout Extensibility Unite rollout.
Software Engineer Intern @ ecobee
- Improved security by removing internal helper fields sent by the public Demand Response API by modifying MySQL queries.
- Implemented PubSub system to broadcast energy surges to thermostats using Golang and Google Cloud, resulting in cost and energy efficiencies.
Software Engineer Intern @ RBC Amplify
- Increased data collection by 56% by developing receipt scanning OCR REST API, grocery list, and onboarding features using React Native, TypeScript, Python, Flask, NodeJS, Express, and PostgreSQL for a Carbon Footprint Tracker Mobile App.
- Collaborated with RBC Ventures’ startup, Goodside, a cross-functional team of 4 interns, and RBC Amplify’s program team.
Software Engineer Intern @ Canada Revenue Agency
- Developed, debugged, tested, and documented projects focused on AI web applications and REST APIs as part of the AI Research team.
- Built a Flask REST API to extract +1000 government benefit reviews from Facebook for a Social Media Sentiment Analysis web app.
- Saved 12 hours/week across 3 teams by developing a tax document translator using Python, Flask, React, and NLP models.
Data Analyst Intern @ Canada Post
- Analyzed 3000+ real estate properties using advanced Excel functions including V-lookup, pivot tables, and macros.
- Wrote Python scripts to automate Excel functions.
- Researched and implemented a $50,000 3D Imaging & VR Facility Scanning Pilot with a team of 6 engineering students.
Softwar Engineer Intern @ Threshold
- Implemented Stripe API using NodeJS to build the backend payment processes for a startup in the event payment space with 1000+ users.
- Developed the frontend checkout page with Stripe Elements, JavaScript, React, and CSS.
Some Extra Activities
Founder @ Studio 50-50
- Founded a design and development shop offering website services to small businesses and startups.
- Raised $10,000 with pitch competitions and grants, selected as one of 30 participants to complete two accelerator programs, the Starter Company Plus and the Summer Company.
- Managed multiple projects, marketed and sold services to an array of clients while building and maintaining good customer relations.
Fellow @ Cansbridge Fellowship
- One of 20 students selected (~4% acceptance rate) to join a global community of leaders, receive $10,000 in funding, partner with Mitacs research organization, and partake in a bootcamp in San Francisco.
Mentor @ Canada Learning Code
- Volenteered as a mentor and assisted in teaching various computer science topics.
Data Associate @ Front Row Ventures
- Conducted due diligence and performed market and company research on student ventures for a student-run venture capital firm.
- Co-led data team initiatives, presented findings to the entire team, created KPI dashboards, wrote Annual Reports and Investment Theses, and increased data collection by implementing a streamlined processing system.
Machine Learning Engineer @ QMIND
- Collaborate with a team of 5 developers and doctors to research, design, and build a Hospital Operation Optimization Recurrent Neural Network model to forecast the daily inflow of patients to the Kingston General Hospital by acuity levels one week in advance with 90% accuracy.
Things I've Built
Featured Project
BONK
A recovery tracker for cyclists that connects to Strava to extract workout data and generate recovery scores and recommendations using a machine learning model.
- Python
- Flask
- MongoDB
- React
- Javascript
- Strava
- Scikit-Learn
Featured Project
Startup Success Predictor
Leverage data science, machine learning, and business principles to identify common characteristics of successful startups to make more informed investment decisions.
- Python
- Flask
- Javascript
- React
- Scikit-Learn
Featured Project
Shiyan Boxer Photography
A website to upload, manage, and showcase my photography through the photo gallery and backend admin panel.
- Javascript
- Tailwind CSS
- AWS S3
- MongoDB
- Python
Other Projects
view the archiveWeb Scraper Deal Sourcing
A web scraper to source businesses' contact information for Studio 50-50's marketing outreach. Contacts are scraped from the web and stored in an Excel sheet. This program was built using Python, BeautifulSoup, Pandas, Requests, and Chrome DevTools.