Lekan (Paul) Soyewo

Product Engineer · Toronto, Canada

lekan.soyewo@icloud.comlekan.cagithub.com/sportynest

Summary

Product-minded software engineer and York University Computer Science graduate. I founded and build PicksHub, a Canadian sports publication, and own every layer of it: the Cloudflare Workers backend, the Next.js website and native iOS and Android apps. Before that I built StudyBuddy, an AI learning platform, at Cognora.

Experience

Founder & Engineer at PicksHub

  • Built and run pickshub.ca, a Canadian sports publication covering news, analysis and predictions across 15+ leagues.
  • Designed the editorial backend on Cloudflare Workers that serves the website, the mobile apps and a Next.js newsroom admin.
  • Shipped a native iOS app on the App Store (SwiftUI, widgets, rich push notifications) and a native Android app (Kotlin, Jetpack Compose).
  • Built PicksHub Play, a predictions game whose results settle automatically from official finals.

Researcher at Cognora

  • Led product vision and development of StudyBuddy, an AI chat platform for learning, from first concept to launch.
  • Engineered the full backend and led a complete redesign of the frontend.
  • Architected an agentic framework that works with all major AI models and providers.
  • Built an internal dashboard for key user metrics, and analysed usage and learning outcomes to refine model routing.

Projects

StudyBuddystuddybuddy.ca

  • Grew an AI-powered learning platform to over 1,000 active users, with real-time collaboration features.
  • Integrated AI models through a custom Node.js framework, improving system performance by 40% with caching and WebSockets.
  • React frontend and Cloudflare Workers backend with sub-100ms response times under high concurrency.

React · Cloudflare Workers · WebSockets · Node.js · Convex · PostgreSQL · Redis

Sentiment Analysis of Movie ReviewsGitHub

  • Built a sentiment classification pipeline on a large movie review dataset, improving accuracy by 15% with TF-IDF and LSTM models.

Python · scikit-learn · TensorFlow · NLP

Moodle Email ScraperGitHub

  • Python scraper with parallel processing via ThreadPoolExecutor, 50% faster than sequential, with rate limiting and error handling.

Python · BeautifulSoup

Education

York University

Bachelor of Science, Computer Science

Dalhousie University

Bachelor of Science, Applied Computer Science

Skills

Languages
TypeScript · JavaScript · Swift · Kotlin · Python · Java · C/C++
Product
Next.js · React · SwiftUI · Jetpack Compose · Cloudflare Workers · Node.js · Convex
AI & Data
LLM APIs · Machine Learning · NLP · TensorFlow · Data Analysis