Creating a machine learning lab for a large financial institution

The client is a world’s leading financial corporation based out of the US with a high analytical maturity in creating and operationalizing analytics projects. One of their largest analytics center of excellence was based out of India consisting of around 1,000 data scientists and has been in operation for over 10 years. The Chief Information Officer (CIO) wanted to enhance their capabilities and experiment with open source platforms and machine learning algorithms (ex: Parallelizing additive machine learning algorithms for Gradient boosting). This meant an increased focus on ‘first of a kind’ experiments targeting niche areas. The organization made the choice of setting up a separate machine learning lab in Bangalore to enable the right culture for experimentation.

The client was facing a number of challenges in talent acquisition, training, process setup, and knowledge management as conventional methods did not cater to creating a “fail fast” cutlure. TheMathCompany is helping the organization to build, operationalize and transfer the machine learning center in a span of two years. As a part of building and operationalizing the center, TheMathCompany did various activities across strategy-execution spectrum.

Design

  • Conduct workshops to discover problem areas and business benefit
  • Interview of stakeholders to understand challenges and individual priorities
  • Working sessions to link the organization vision to the machine learning lab
  • Creation of a pipeline of futuristic experiments to set industry benchmark

Enable

  • Development of modularized training programs to cater to dynamic needs
  • Setup of multiple delivery modes to ensure continual and easily accessible learning
  • Identification of the right tech stack for a higher end lab
  • Design and execution of hackathons to tap into a smarter talent pool
  • Introduction of the ‘right’ budding start-ups working on niche area of analytics like AI and Cognitive science
  • Scaling of the team organically with proven successful talent profiles
  • Setup of an ‘Analytics incubator’ enabling internal, external enthusiasts to experiment
  • Setup of knowledge management frameworks to capture results of experiments for reuse