Skip to content

hi, i'm —

CalvinLiew.

I ship products at the intersection of AI agents, data, and product strategy.

Right now I'm leading SupRM Intelligence at Sanofi, an AI agent that turns scattered supplier data into real-time risk, performance, and QBR insight for 30+ stakeholders.

Featured projects(my favorites)

Recent work I'm proud of, across AI agents, product, and data.

AI & Data

SaaSScout: A Grounded RAG Copilot for SaaS Evaluation

Independent Project

A production RAG copilot for SaaS evaluation. Indexes 335 products and 4,899 review chunks into four partitioned Chroma vector collections, ranks retrieval candidates across six signals (feature-fit, pricing, review sentiment, provenance trust, query alignment, category overlap), then generates grounded recommendations via a provider-neutral LLM layer — Groq Qwen 32B online, Ollama Qwen 2.5 local, or a deterministic grounded template when both are unavailable. The architecture ensures the LLM never invents a fact the retrieval layer did not supply.

RAGFastAPIReact+12 more
AI & Data

The Anatomy of Fear: Quantifying Horror

University of Toronto · CSC316H1

An interactive data visualization study of how horror films generate fear. Analyzed 129 horror screenplays across 9,760 scenes and 11,204 horror signals using a GPT-4o-mini + GPT-4o pipeline (99.97% success rate, $2.28 total cost), then built nine custom D3.js visualizations into a scrollytelling site that reveals horror's power law: a tiny set of elite signals like scream and blood drive most of the actual scares.

LLMGenerative AID3.js+4 more
Product

Drive Companion: AI-Powered Cloud Intelligence

University of Toronto · MGMC11H3

A high-fidelity prototype reimagining Google Drive as a proactive AI partner instead of passive storage. Six signature scenarios (Syllabus-to-Schedule, Smart Reading Pack, Living Concept Maps, AI Meeting Chief of Staff, Executive Briefs, Work Rhythm Optimiser) demonstrate how an embedded AI layer can replace the manual-upload chore that defines ChatGPT and Copilot. User research with 100+ students and professionals exposed the real lever: emotional benefits (anxiety down, confidence up) lift adoption far more than time savings alone.

Product ManagementProduct DesignPrototyping+4 more

What I do(the toolkit)

What I bring to AI product work. Designed to ladder from raw data  →  shipped agent.

ai & data.

  • AI Agent Development
  • LLM Orchestration
  • Prompt Engineering
  • Large Language Models
  • Snowflake
  • API Integration
  • Data Pipelines
  • Data Analysis
  • Business Intelligence
  • Data Visualization

product.

  • AI Product Strategy
  • Product Development
  • Requirements Analysis
  • Stakeholder Management
  • User Research
  • Agile/Scrum
  • Roadmap Planning
  • Product Analytics
  • A/B Testing

technical.

  • Python
  • JavaScript/TypeScript
  • React/Next.js
  • SQL
  • REST APIs
  • FastAPI
  • Git
  • Low-Code Platforms

design.

  • UI/UX Design
  • Prototyping
  • Figma
  • Design Systems
  • Wireframing
  • Interaction Design

Drop me a line.

always up for a chat about ai, product, or design

Hiring, collaborating, or just want to chat? I'd love to hear from you.