Portfolio

— Selected projects

1. End-to-End Recommender System — Tastet Project

— Designed and deployed a hybrid recommender system on GCP, benchmarking 8 ranking models with a dual-metric evaluation (F1-score + custom fairness metric). Event-driven ML pipeline with GCP Cloud Functions and weekly cron-based retraining.

— Mitacs-funded ($15,000+), supervised research at HEC Montreal with Prof. Laurent Charlin.

— Learn more about this → here

Tastet Recommender System

2. Product Recommendation Engine — Browns Shoes

— Engineered a recommendation engine leveraging order history patterns to personalize product grids in customer communications.

Product Recommendation Engine

3. Olima.ai — Similarity-Based Recommendation App

— Built a similarity-based recommendation app using Wikipedia clickstream co-occurrence data and spectral embeddings. Celebrity similarity model deployed as a mobile app (React Native / Expo).

olima.ai

Olima - Your crowd reveals who you are

4. AI-powered Quality Assurance Tool (GPT + HuggingFace)

— Prototyped an AI system to detect misleading product descriptions on e-commerce platforms using GPT and HuggingFace.

AI Quality Assurance Tool

5. Text Classifier GPT Tool

— Built a GPT-4 powered tool for bulk text classification, deployed as a Streamlit app.

6. Customer Scoring System

— Predictive models for customer lifetime value, churn prevention, and engagement scoring.

7. Vibecope.ai — Get the Best Out of Claude

— Built a macOS dev tool using event hooks for real-time AI prompt coaching. Reads Claude sessions and injects coaching into context.

vibecope.ai

Vibecope - watches you work with Claude

8. Data Reporting & Visualization

— Interactive dashboards and automated reporting infrastructure.