This paper introduces a novel dataset aimed at advancing stress detection in software developers through the correlation of electrodermal activity (EDA) and heart rate variability (HRV) with occupational stress. Our primary objective is to deliver a comprehensive, multimodal dataset to empower future machine learning research, particularly for training algorithms like artificial neural networks (ANNs) to identify stress in this high-risk population. Data were collected from sixteen software developers during a controlled experiment featuring relaxation and stress-inducing tasks (Stroop and arithmetic tests), captured via the Emotibit wearable device, and supplemented by Perceived Stress Scale (PSS-14) scores to contextualize participants’ stress levels. Analysis revealed distinct physiological patterns: group-level data showed modest heart rate (HR) and EDA increases, with reductions in HRV metrics (RMSSD, SDNN, pNN50) under stress. The individual analysis identified significant EDA increases in nine participants, with subgroup HRV reductions (SDNN: p = 0.0130; pNN50: p = 0.0115) underscoring personalized stress responses. These findings affirm the dataset’s robustness for developing tailored stress detection models. The dataset stands as the core contribution of this work, offering a valuable resource for machine learning-driven stress monitoring systems. By targeting software developers, it paves the way for innovative solutions to enhance well-being and productivity in this critical workforce.