Research Summary: Detecting Soft Skills in ML Engineering Roles CVs

Research from arXiv explores how candidates articulate soft skills in ML engineering roles through their CVs, contrasting this 'supply side' perspective with traditional 'demand side' views from job advertisements and employer surveys. The study addresses the limitations of keyword-based CV analysis by employing an LLM-based approach to extract both explicitly listed and implicitly narrated soft skills from a corpus of 300 curated CVs across ML engineers, data scientists, and software engineers.

Why it matters

Understanding how candidates present their soft skills is critical for talent acquisition strategies, enabling organizations to refine job descriptions and interview processes. This insight helps align candidate self-presentation with organizational needs, fostering more effective team collaboration and project success in technology-driven environments.

Key insights

• Existing knowledge on soft skills in ML engineering largely stems from employer demands (job ads, surveys, hiring manager interviews).
• Candidate articulation of soft skills in CVs has been under-researched.
• Previous CV-mining methods for soft skills were keyword-based, missing narrative expressions.
• Prior descriptive studies reported frequency rankings without statistical validation against sampling variation.
• A new LLM-based method extracts both explicit and implicit soft skills from CV narratives.
• The study uses a balanced corpus of 300 CVs spanning ML engineering, data science, and software engineering roles.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.10046