Solution for Predicting Employee Performance and Department Fit Using ML/NLP Linguistic Analysis
HR-Tech/Banking
North Carolina, USA 
6 months
Custom Solution Development|Data Engineering|NLP and Machine Learning Integration
IBM Watson|ML|Node.js|Python (SciPy)
Dedicated team behind the project
The Client
The client is a company who provides data engineering services for the largest multinational investment banks with headquarters in North Carolina, having a strong presence in the banking sector. They have an R&D office that is involved in the development of SaaS products for automation and optimization of business processes around those industries. These solutions are initially deployed across their branches for better performance and efficiency; later they expand to businesses of other industries such as sports, construction, among others, delivering them advanced workforce optimization tools.
Client Achievements:
The Challenge
The customer wanted to reduce employee turnover caused by poor fit at the department or role. They realized that a data-driven selection could strongly enhance their hiring process by identifying candidates’ strengths and aligning them with suitable departments. The core challenge was to develop a system capable of analyzing linguistic and emotional traits in written responses, predicting not only performance but also long-term department compatibility. The idea was to implement this proof of concept in a pilot branch before scaling it across various industries.
What Was Done
We developed a high-level natural language processing (NLP) and machine learning (ML)-based solution to analyze written employee responses. This system used IBM Watson’s Tone Analyzer API to capture emotional insights, which were then processed through a custom algorithm to suggest ideal department matches. The initial proof of concept targeted one branch of the bank. Based on its success, the solution was scaled to other industries, including construction and sports, proving the approach was versatile and flexible.
The Model Workflow​
Implemented Features:
We designed an engine that uses NLP to analyze written text from employees and candidates, generating performance predictions based on linguistic and emotional traits.
Emotional analysis was implemented using IBM Watson’s API, identifying key emotions such as joy, anger, and satisfaction, which were fundamental in predicting department fit.
A custom algorithm was created using Python (SciPy) and R to process emotional data and determine department suitability, offering predictive insights specific to the bank’s needs.
Google Forms was used for survey collection, while Google Sheets allowed real-time processing and score adjustments, allowing the data insights to be dynamically updated.
Custom-built statistical models analyzed linguistic data to assign performance scores, offering a quantitative approach to predicting employee success within departments.
We provided automated reporting tools that generated detailed visual representations of employee performance and department matches, making it easier for managers to interpret and act on insights.
Employee Performance Ranking
Results
After 6 months of analysis, the solution reduced employee turnover to 18%, a significant improvement from the industry average of 23.4%
Better department matches led to improved retention rates by aligning employees’ strengths with the right roles
The model was successfully tested on 50 new hires. The results showed a high correlation between the system’s predictions and actual employee performance, demonstrating the tool's effectiveness in hiring decisions
Over 400 existing employees were analyzed to create a benchmark for future hiring decisions. This allowed the client to set a standard for future hires, ensuring new employees met or exceeded existing performance metrics
The solution was implemented across the banking, sports, and construction sectors. Each industry had a different candidate pipeline and headcount, but the system consistently delivered valuable insights for employee placement and performance prediction
Tech Stack
Daniel
Head of Engineering at Ralabs
Predict Employee Success with AI-Driven Insights
Our AI solutions identify the best-fit candidates for your team, reducing turnover and enhancing employee performance from day one.
Other Cases
