Aligning individual passions with professional capabilities remains a critical challenge in the modern labor market. Traditional career guidance often relies on resume data, which suffers from bias and low predictive validity. This thesis investigates whether a Hybrid Job Recommendation System, integrating subjective task preferences (Needs-Supplies Fit) and objective cognitive profiles (Demands-Abilities Fit), improves the perceived relevance of career guidance compared to single-modality models. A “Unified Task Universe” was constructed to harmonize O*NET, ESCO, and SSG occupational frameworks. Then, it was used to build an interactive dashboard where participants could rate activities and get job recommendations. In a within-subjects experimental design, 40 professionals (N = 40) evaluated job recommendations generated by four distinct models: Task-Only, Cognitive-Only, Hybrid, and a Random Baseline. Results from a Friedman test revealed significant differences among the models. Contrary to the central hypothesis, the Hybrid model did not yield the highest user-rated relevance. Instead, the Task-Only model significantly outperformed all others, indicating that users prioritize subjective interest over objective capability. However, a validity analysis revealed a critical paradox: while the Task-Only model lacked criterion validity, the Cognitive and Hybrid models demonstrated validity coefficients (ρ = 0.24) double that of traditional CV screening. These findings suggest that while cognitive data successfully identifies hidden potential, users often fail to recognize these unfamiliar matches as relevant. Future systems should employ Explainable AI to bridge the gap between what users want to do and what they are objectively capable of doing.