Miniconf on Cloud Server Management (New Delhi)

This talk encompasses is a narrative of our journey from manual releases to an entire CI pipeline that allows us to release code hourly. How the CI Pipeline was modified to help move data and machine learning models across environments and the learnings along the way.

We have empowered our team to deploy code several times a day, as well as their chat flows and data. It encompasses a pipeline for both code deployments as well as ETL, tagging and data movement for machine learning models.

This talk is not a deep dive into technicals of Jenkins, Ansible etc but a high level overview of system, the thought process and the business value derived from the system.

Outline

The talk is presented as a narrative of our journey. Our initial architecture that involved manual releases and the challenges therewith. The challenges in building a testing framework for chat flows and the evolution of the full pipeline.

Note: The talk in the video given earlier is not up to date with improvements we have made to the machine learning data pipeline and the data tagging.

Speaker bio

Prateek Gupte is currently the Director of Engineering at Haptik, India’s first conversational commerce platform. He has held engineering leadership positions at his previous startups - BYOF Studios and Code Red, and has built technology solutions for a variety of domains such as gaming, healthcare, pricing solutions and artificial intelligence.

Slides

https://www.slideshare.net/PrateekGupte/continuous-integration-haptik