Senior Computer Vision Engineer (AI)

Location: Sant Cugat del Vallès, Spain — Hybrid

Company

hirify.global develops mathematics-learning resources, training, and research for schools and teachers; its Mathboard AI technology interprets students’ handwritten pen strokes.

What you will do

• Own the architecture, representation, training, assignment, and improvement of handwriting-reading models.
• Design, implement, and evaluate new approaches for multiple-instance detection over geometric pen-stroke primitives.
• Run controlled comparisons with predefined success criteria, analyze regressions and trade-offs, and document unsuccessful experiments.
• Deploy models to ordinary school hardware with low latency, deterministic behavior, evaluation parity, and rollback capability.
• Monitor live performance on student work and use observed behavior to define subsequent experiments.
• Collaborate with dataset and evaluation owners to distinguish model, data, and task limitations.

Requirements

• Deep experience with multiple-instance detection or instance segmentation, including prediction matching and metric interpretation.
• Experience with non-rasterized inputs such as strokes, polylines, trajectories, point clouds, graphs, vector graphics, CAD geometry, or sensor traces.
• Knowledge of attention-based architectures over sets and graphs, including spatial and relational structure encoding.
• Serious hands-on experience training deep models in PyTorch, including training loops, schedules, numerical stability, distributed runs, and debugging.
• Experimental discipline with fair comparisons, reproducibility, and critical evaluation of research literature.
• Production experience covering export, latency, determinism, and rollback paths.

Nice to have

• Online handwriting, sketch, ink, or diagram recognition experience.
• Geometric deep learning or graph neural networks for spatial data.
• Vector graphics, CAD, GIS, or trajectory modeling experience.
• Inference optimization, quantization, or published/open-source work.
• Spanish or Catalan.

Culture & Benefits

• Small technical team with direct access to product decision-makers and substantial model ownership.
• Experiments and results are documented, including failed attempts.
• Results are assessed against criteria agreed before each run, with data checked before reporting metrics.
• Access to the GPUs required for model development.
• Work is expected to progress from research engineering into classroom use.