Context

Xccelerate, a pioneering train-to-hire platform, had a goal of training 100 of the most high-potential candidates for Solana designers and engineers on behalf of the Solana Foundation from amongst the many applications they received for their selective “train-to-hire” course. The Xccelerate team needed a solution to streamline their candidate selection process for this selective course while maintaining their high quality standards, and brought in Axiom to develop an innovative AI-powered online assessment tool.

Challenge

Previously, Xccelerate faced challenges with attracting and admitting the best candidates for their train-to-hire programs. Unmotivated or unqualified candidates dropping out partway through the courses, or failed to pass the final exams, which posed challenges and created reputational risks for Xccelerate. Even when candidates successfully graduated with the required technical “hard” skills, they didn’t always bring accompanying soft skills (such as cooperation, interpersonal skills, critical thinking, etc.) needed to flourish in their roles. Xccelerate had long wanted to test for soft skills, but employing the dedicated human assessors needed to read the long-form answers of hundreds of course applicants would have been prohibitively costly, so the team was at a loss.

Solution

Axiom proposed a comprehensive AI-powered assessment tool designed to evaluate train-to-hire course candidates’ “hard” and “soft” skills. The AI-powered assessment engine is built atop Typeform, an existing survey platform, where questions are asked in a way that simulates an actual job interview and allows candidates to provide long-form answers and showcase their previous work.

Implementation

The technology stack consisted of:

• Front-end platform: A Typeform survey that formed the user input for our frontend.

• Back-end architecture: Inputs from the Typeform were fed into our easily modifiable and updatable backend, built using Vercel and containing the guiding prompts for ChatGPT.

  • We employed ChatGPT in our backend to evaluate the short- and long-form qualitative responses from the front-end Typeform survey, which included descriptions of students’ past projects, scenario-based questions about team dynamics, and other questions exploring candidates’ soft skills.

• Student-facing output: The tool would provide a detailed feedback report to the candidates through email, offering them insights and recommendations.

• Internal-facing assessment: The tool sends an assessment of students’ current skill levels to the Xccelerate team, where they are visualized alongside other students’ scores to allow for simple comparison across areas.

• Currently in development: Integration of image recognition tools to assess design portfolios for originality, attention to detail, consistent visual layout, and other qualities.