Description and Outcome: AiDa Develop is a platform which enables seamless experience for machine learning model development including all major stages like hyperparameter tunning, training, model interpretability, performance evaluation and data quality. Platform owns range of products implemented and optimized to support these various stages smoothly and efficiently for 1000s of modeling team members.

In the project, candidate (intern) will be working towards analyzing the current logging framework and identify the opportunities, come up with the right recommendations either to streamline or enhance or build something new from scratch. Project will also include building a POC (proof of concept) with following scope.

1. Incorporate recommendations in the current logging framework and upgrade to be comprehensive and intuitive

2. Build a systematic solution (pipeline) to generate insights (some examples below but not limited to) for leadership to take a judicious decisions towards solving users pain points proactively or enhance user experience even more.

Example insights – top n failures at month/quarter level, top products facing maximum or minimum failures. Highly utilized or under-utilized products. Average, min, max time taken by various stages

Skill sets: Python, BIgQuery (GCP cloud), Understanding of Unix, understanding of model development life cycle and various stages and their objective like hyper parameter tunning, training, model interpretability.

Expected learning: Understanding of model development life cycle and practices being followed by Amex, Big Data Environment and execution mechanism for a distributed job. Google cloud framework Principles of optimizing the Big Data Jobs on cloud. Generate business insights using programming methods.

Specific courses required for project execution: Programming in Python, Linux / Unix understanding, Big Data framework such as Big Query, DataProc and other GCP essentials (Google Cloud Platform), Algorithms, classical ML algorithms such as XGBoost in addition to Linear Regression, SVMs, other classification / regression methods.

American Express - AiDa, Bengaluru

Station Overview

Project 1

Title: Streamline and enhance model development logging framework and build a systematic analytical solution to enable insights like top n user concerns, highly used or unused products etc