Arun Thiru

Data Science, Fitness, Thoughts

My Headshot

About

I'm a Senior Data Scientist at Eyeo GmbH (they own AdBlock and AdBlock plus). My work interests include Building algorithms for Real Time Bidding (RTB), Time Series Forecasting and Causal Inference. I also enjoy writing, reading and fitness.

Skills

Time Series Forecasting
Causal Inference
Optimisation Algorithms
Classical Machine Learning
Transfer Learning
Data Vizualisation
ETL pipelines
Machine Learning Systems Engineering
Data Analysis

Work Experience

Blockthrough (acquired by Eyeo GmbH)
Remote, Canada

June 2022 - Present

Senior Data Scientist

    + Leading the development of our Price Floor and SPO model to optimize revenue. I have explored and combined various strategies using traditional economics based models to building machine learning models to predict and set the optimal price floor for each ad unit.
    + Implemented several Causal Machine Learning methods to run 'what if' analysis on several business decisions across various clients and geos.
    + Created a Hierarchical Revenue Forecasting Model to forecast revenue monthly for the next 12 months and also a daily forecasting model to identify when clients are underperforming so we can proactively address the issues
    + Developed a tool to automate A/B testing and product analytics
    + Built several ETL pipelines to create dashboards and reports for the company
    + Mentored and trained junior data scientists and data analysts
    + Lead team whilst director was away on Maternity Leave

Royal Bank of Canada (RBC)
Toronto, Canada

April 2019 - May 2022

Data Scientist

    + I developed the nationwide cash forecasting system to predict how much cash a branch would need to order on a daily basis to meet customer demand. This system was deployed to the entire country saving $2 Million (CAD) annually for the lifetime of the project.
    + I built a Credit Recommendation Model which cross recommended credit products to our existing small business clients. The model resulted in 55 new products and 170 accounts with a total net limit increase of 12MM in just Q1 2021/22
    + During the initial weeks of covid I was tasked to develop a way of showing how transaction activity had changed across the Greater Toronto Area (GTA). We used this to inform policy on branch closures and cash management This dashboard was presented to the vector institute.
    + Created the MVP model for the RBC personalization product. The product was designed to give personalized offers on the RBC avion platform.

Royal Bank of Canada (RBC)
London, UK

June 2018 - September 2018

Data Science Intern

    + Amplify is an international program for teams of students to build disruptive technology solutions for business problems within RBC.
    + We built RBC Ark a tool which used a combination of supervised and unsupervised machine learning techniques to create a novel hashing algorithm to match client information from disparate databases. We co-authored a provisional patent for this algorithm.
    + We presented our solution to the RBC CTO and other senior executives beating out 18 teams to win the 'Most Disruptive Technological' solution and $10,000 in prize money.