September 27, 2019
The work to package and deploy machine learning/AI is often not as celebrated as the ML models themselves. Here at gk, we pride ourselves on having an excellent engineering team.
Here are some September product updates from our Platform/Engineering team.
Migration from Python to SQL
Our chat analytics’ first version were coded in a very popular Python data analysis library called Pandas. Pandas is a great tool for exploring and manipulating data, but its usability comes at the cost of speed. To scale our results and process large datasets, we needed be able to analyze large chat files much faster. After rewriting all the queries in SQL (which our database Postgres is optimized), the processing time was sped up 11x to generate conversation insights.
Processing Chat Data 50X Faster!
The work to package and deploy machine learning/AI is often not as celebrated as the ML models themselves. Here at gk, we pride ourselves on having an excellent engineering team that builds seamless deployments and is constantly focused on optimizing processes to run faster. This month we released a new XML parser that is 50X faster than our previous version. This means we are able to process and return insights on YEARS of chat history data for our clients – not just days or weeks.
Get Deployed in 1 Day with Kubernetes!
At gk, the platform team puts a lot of thought behind making rollouts of our NLP engine seamless, which can be deployed in under “a day”.
This month, to improve the scalability and reliability of our deployment packages, the team containerized gk’ NLP applications on the best in class container application, Kubernetes. Kubernetes (K8s) is open-source system for automating deployment, scaling, and management of containerized applications, which was originally designed by Google.