Increasing Placement Revenue by Scaling from 5 to 1000+ Users
Overview
A large recruiting and staffing enterprise relied on a legacy Machine Learning algorithm for candidate matching. The existing offline model limited scalability and performance, constraining growth and future initiatives. ImagineX partnered with the client to modernize the ML system, improve model effectiveness, and implement a roadmap for sustainable, enterprise-scale operations
Problem
Existing ML algorithm required offline training, limiting scalability.
System supported only 5 users, impeding business growth.
Model performance needed improvement to increase prediction accuracy.
Data quality issues impacted ML effectiveness.
Solution
ImagineX analyzed system bottlenecks, defined business objectives, and designed an improved architecture. A Serverless approach in Databricks replaced the Kubernetes cluster, enhancing scalability. Data and feature engineering were optimized, ML best practices were implemented, and a detailed roadmap was developed outlining architecture, processes, and skill requirements for more accurate and efficient operations.
Outcome
Scaled solution from 5 users to over 1,000 users.
Improved ML model effectiveness, increasing accuracy of candidate matching.
Enhanced data quality and feature engineering for reliable predictions.
Delivered a comprehensive ML roadmap covering architecture, process, and team requirements.
Educated client teams on sustainable ML operations best practices.
Data & AI Engineering
MLOps Architecture Roadmap
Data & Feature Engineering
Technology & System Architecture
MLOps Engineering
Technology
Microsoft Azure
Databricks
