Descripción de la oferta
Job SummaryThe successful candidate will join the Germans Trias i Pujol Research Institute (IGTP) as part of a multidisciplinary team developing AI‑based computer vision systems for surgical instrument recognition. The project is carried out in collaboration with HealthTech Innovations and is deployed on a robotic platform used in the central sterilization unit and operating rooms of the Hospital Universitari Germans Trias i Pujol. The role focuses on the design, training and optimization of computer vision and deep learning algorithms, covering the full pipeline from data acquisition to model deployment. Technical autonomy is required, with responsibility for code, model iterations, architecture decisions and integration with the robotic platform.Main ResponsibilitiesLead the development of the AI‑based surgical instrument recognition system, with direct responsibility for design, training and performance of its computer vision and machine learning components.Design and develop computer vision pipelines for surgical instrument detection, classification and tracking, covering the full lifecycle from data acquisition and preprocessing to model training, evaluation and deployment.Write, optimize and maintain production‑quality code, ensuring robustness, scalability and reproducibility of AI models.Train and fine‑tune machine learning and deep learning models, selecting appropriate architectures, defining evaluation metrics and implementing strategies to improve accuracy and generalization in real hospital environments.Generate datasets, define labeling strategies and data augmentation techniques, and iterate on models based on validation results.Collaborate with the robotics engineering team to support integration of the AI/computer vision modules into the robotic platform, focusing on the perception side of the perception‑action loop.Test and validate the system in the central sterilization unit and operating room, analyze performance, identify limitations and propose technical improvements.Document algorithms, code and experimental results to support regulatory processes and scientific dissemination.Qualifications and ExperienceDegree in Computer Science, Telecommunications Engineering, Data Science, Artificial Intelligence, Mathematics/Physics with a computational focus, or a related technical field.Master’s degree or specialized training in Artificial Intelligence, Machine Learning, Deep Learning or Computer Vision; or equivalent professional experience.Strong hands‑on experience developing computer vision and deep learning systems for real‑world applications (object detection, segmentation, classification, tracking).Proficiency in Python and practical experience in at least one major deep learning framework (PyTorch or TensorFlow).Experience with OpenCV and the full model lifecycle—dataset generation, labeling, augmentation, training, evaluation and iterative optimization.Comfortable operating in multidisciplinary environments, with the ability to work closely with clinical staff.Fluency in Spanish (mandatory); Catalan valued as an asset. Good level of English for technical documentation and collaboration.Strong problem‑solving skills, autonomy, proactive attitude and genuine ownership of technical decisions.Availability to commute regularly to the Hospital Universitari Germans Trias i Pujol or to the HealthTech Innovations offices.Additional Qualifications (Optional)Knowledge of C/C++ and robotics middleware (e.g., ROS).Prior exposure to robotic perception systems or integration of AI components into a robotic platform.BenefitsIndefinite contract under science law linked to Project CPP2024‑011704, fully funded by MICIU/AEI/FEDER and the EU.Immediate incorporation into a highly innovative applied research and development project with direct impact on real clinical environments.Full‑time position with a 39‑hour work week and flexible working schedule.Hybrid model allowing partial remote work with regular on‑site collaboration sessions.23 days of paid annual leave, plus 5 additional days for personal matters, 2 days for Easter and 2 days for Christmas.
#J-18808-Ljbffr
