Pinned Loading

1. Movies-to-Book-Recommender-Cold-Start- Movies-to-Book-Recommender-Cold-Start- Public

   Semantic Bridges: LLM-Driven Cross-Domain Cold-Start Recommendation (Movies → Books). Built a movies-to-books recommendation engine for cold-start discovery by comparing zero-shot semantic retrieva…

   Jupyter Notebook

2. computervision-instagram-engagement-analysis computervision-instagram-engagement-analysis Public

   End-to-end unstructured data pipeline that scrapes @purenewzealand Instagram posts, extracts computer-vision labels, and models which visual + caption themes drive engagement.

   Jupyter Notebook

3. snowflake_streamlit_reporting snowflake_streamlit_reporting Public

   Built a Snowflake bronze-to-silver analytics pipeline on Amazon product catalog, automating ingestion with stored procedures while publishing reusable KPI views and Streamlit dashboards for busines…

   Jupyter Notebook

4. custom-word2vec-crowdsourced-recommender custom-word2vec-crowdsourced-recommender Public

   Crowdsourced beer recommender built from BeerAdvocate review text—comparing TF-IDF bag-of-words, spaCy embeddings, and custom Word2Vec—enhanced with sentiment-aware scoring and competitor discovery.

   Jupyter Notebook

5. lp-mip-optimization-gurobi lp-mip-optimization-gurobi Public

   Optimized a $10M multi-channel marketing budget using linear and integer optimization with tier-based ROI curves, maximizing return under real-world constraints like spend caps and channel ratios.

   Jupyter Notebook 1

6. nlp-driven-competitive-benchmarking nlp-driven-competitive-benchmarking Public

   Built a competitive positioning map for the entry-level luxury car market by mining ~$10,000 forum posts and applying frequency-lift + multidimensional scaling, translating unstructured demand sign…

   Jupyter Notebook 1