Lynnette Tee

AI-powered content strategy platform helping expert business owners grow through short-form video

• Own product management and roadmap prioritization across the platform
• Design and evaluate the LLM prompts behind content strategy and video ideation features
• Write specifications and coordinate feature development with engineering
• Run go-to-market experiments and synthesize customer feedback into product decisions

Kuala Lumpur, Malaysia

• Managed campaign budgets across multiple categories and quarterly initiatives, improving ROI through cost-to-income ratio analysis
• Built and maintained over a dozen dashboards in SQL, Python and Google Sheets to track category spend and financial KPIs
• Redesigned the protocol for recurring high-value seller payments, cutting processing time by 400% and reducing manual error with Apps Script automation
• Led financial planning for the platform-wide 2024 Ramadan campaign in fashion and beauty, coordinating business development, marketing and tech over a 14-day sales cycle with zero budget overrun

Markham, Ontario, Canada

• Worked on watsonx Orchestrate, an AI agent orchestrator that automates business workflows
• Cut CI/CD pipeline runtime by 70% and expanded product reach by shipping a Slack integration
• Debugged 60%+ of a production ML recommendation pipeline, fixing 2 critical-severity bugs within 2 days
• Integrated services with legacy systems through Java and JavaScript REST APIs; ran quarterly security scans on released microservices
• Led a 5-person IBM Jumpstart team that turned customer voicemails into Salesforce tickets with Watson NLP and Speech to Text, raising answer rates by 90%

Full-stack app that surfaces friends' Instagram stories without the doomscroll. Containerized ML backend uses a vision-language model to read story context; React and TypeScript frontend.

End-to-end photo pipeline using shell scripting, Lua and a vision-language model to transfer files, assess image quality and auto-tag shots.

Web app built in a one-day hackathon that generates personalized, narrated children's stories with an LLM and text-to-speech.

Capstone with a team of 4: a hybrid NLP pipeline of rule-based heuristics, K-means clustering on BERT embeddings and WordNet to generate and compare ontology graphs from unstructured text.

Composite model of logistic regression, decision trees and random forests reaching 88.6% accuracy; ranked 1st of 26 class projects.

BASc, Industrial Engineering · Minors in Artificial Intelligence and Engineering Business

Runner-up, U of T Re-Engineering Competition (2019)