STEP uses temporal and educational signals for career-path recommendation

The authors present STEP (Sequential Trajectory of Employment Prediction), a novel career-path recommendation system that leverages temporal and educational signals from unstructured resumes to predict the next job in a career trajectory. To improve internal occupation representation, the system introduces ROUTE, a two-stage contrastive procedure that adapts a multilingual encoder via unsupervised denoising autoencoding followed by supervised contrastive fine-tuning.

• STEP integrates a time-decay Gated Recurrent Unit (GRU) cell to model temporal dynamics.
• It uses Feature-wise Linear Modulation (FiLM) conditioned on educational attainment.
• The system employs attention-based sequence pooling to select relevant features for next job prediction.
• Evaluation on four datasets, including an improved JobHop dataset, shows STEP outperforms state-of-the-art baselines in next job prediction.

The dataset and code are publicly released to support reproducible career-trajectory research.