● Built and deployed Marketing Mixture Models using LMMM (in Python) & Robyn (in R) to help optimize marketing
spending strategies in GCP and Databricks; displayed results in Looker visualizations for business stakeholders.
● Planned and conducted Incrementality testing strategies using Geolift in R to evaluate the true impacts of a given channel.

● Computer Vision Models: Helped the client add a gaming layer in tennis live streams to improve fan engagement. Tested different ways to strengthen the game engine using computer vision. Self-learned, created, and tuned deep learning model
prototypes using tensorflow/keras and pytorch to identify scoreboard changes and camera shifts for game system resets.
● Recommendation Engine: Mined user data (3M+ users) and product features (80+ products) in Snowflake and Databricks, and
created ML models/recommendation engines to improve customer experiences and engagement when browsing products.
● Data Pipeline Creation: Created a data pipeline in Python for Ad Spend Optimizations on the client’s acquired brands on the Walmart website. Web-scraped product reviews from the Walmart website daily using AWS Lambda and ingested data into the
PostgreSQL database; presented outcomes on Tableau and triggered alerts based on the ad spend requirements.
● Network Analysis: Processed thousands of browsing data from Yelp users using pandas and constructed network graphs using
networkX, pyvis to find which restaurants are viewed together more often, to build a recommendation system.
● Retool BI Dashboards: Created multiple dashboards in Retool to monitor daily KPI status by extracting and aggregating data in
MySQL. Connected a “feedback” dashboard with the database through a webhook for team leads to monitor real-time feedback.

● Created data pipeline for lead delivery to the company’s 3rd-party mortgage agents using Pyspark and AWS Glue; tested API calls
using Postman; conducted A/B testing to compare the conversion between the experiment and control groups.
● Generated $55 million extra closing volume in the first 4 months of implementation.

● Improved the efficiency and accuracy of the underwriting process by solely initiating and completing 3 end-to-end ML products
for evaluating and predicting customers’ engagement and credibility.
● Worked with data from 10+ sources and built executable functions that contained data preparation, model building & selections
in Python and R; monitored model performances using DOMO; set up the data and ML pipeline using Jenkins.
● Saved $20k per month and greatly improved major KPIs (conversion%, default%) by optimizing customer selection strategies.

This is the website of my personal consulting firm. It contains the project areas that I can help with and a list of happy clients I have partnered with. Check it out!