I am a Data Science professional having 13 years of experience in the IT industry. I am currently working as a Sr. Data Scientist with Autodesk. Before joining the PGP-AIML program, I was a Data Science Professional.
The customer support team at Autodesk receives many application errors reports wherein the team manually verifies those reports and routes them to the respective teams/organizations to solve the issues. This is an unsupervised scenario where we have applied the ML technique to read through the context of the error and categorize the error data (Customer Error Report)
Every application/product of Autodesk generates many Customer Error Reports ( these are the error log files), which are stored in the AWS cloud platform. Our Data Science team extracts these reports( text files), extracts required data from those error reports, cleans the data, applies text data cleaning methods, and uses the K-means Clustering algorithm to cluster the reports.
This delays solving the customer issue and requires manual effort to route the issue to the right team.
We have used the Python tool NLTK to clean the text data and the K-Means clustering algorithm to cluster the reports. And AWS lambda to deploy the model.
Over the period in the AI/ML course, I learned many techniques to solve some of the challenging business problems. So I found the opportunity to apply ML techniques like clustering/ classification (if we have supervised data) in this particular business problem to solve the issue, and it happened.
We have recommended the solution to cluster the error reports in different segments and route them to the right team quickly to take immediate action.
After implementing this solution, we have increased customer efficiency and decreased the turnaround time to solve the error issue. Also reduced manpower which saves 3FTE annually.
This was a great project I worked on where I have implemented my learnings from the AIML course. Still, there are many things to learn, and I will be using my learning to solve many more business problems.
