An in-house platform for conducting A/B experiments can be a valuable tool for any company that is looking to optimize its online offerings, such as a website or a mobile app. This talk will cover the creation of an in-house platform for conducting thousands of parallel A/B tests. The talk will discuss the key considerations and challenges in setting up such a platform, including experiment design and analysis and best practices for implementing and scaling an A/B testing program and utilizing the results of the experiments.
Key Takeaways
- Introduction to A/B testing & how it can be used to improve business metrics via data-driven decisions
- How to design the A/B system that allows us to conduct thousands of parallel experiments?
- How to dynamically split the traffic so users can be assigned to experiments?
- How to assign users to variants so the audience is homogenous
- Introduction to statistical significance calculation to validate the A/B experiment results
————————————————————————————————————————————————————
Bio
Rama Badrinath – Principal Data Scientist | Meesho | India
Rama Badrinath is a seasoned data scientist with extensive experience in machine learning. As a Principal Data Scientist at Meesho, one of India’s leading e-commerce companies with a valuation of $4.9 billion, he is responsible for driving the company’s data-driven strategy. Prior to his current role, Rama held positions at Microsoft and Sharechat. He holds a Master’s degree in Machine Learning from the Indian Institute of Science (IISc) and has published papers in prestigious conferences such as SIGKDD and ECIR.
ANZ-Stage 2
Rama Badrinath – Principal Data Scientist | Meesho | India