This position requires a highly motivated person who wants to help us advance the understanding of multilingual capabilities of multimodal foundation models. You will be responsible for designing and running large-scale experiments, and for creating specialized training and evaluation datasets to study how multimodal foundation models acquire multilingual capabilities during training, as well as to devise training strategies that facilitate optimal multilingual transfer. The adecuado candidate combines deep technical expertise in machine learning with a proven track record in multilinguality and/or multimodality, the ability to write high-quality code, work with large-scale systems and solve challenging problems independently. Strong interpersonal and collaboration skills are also essential.
- Currently working towards a PhD degree in Computer Science, Machine Learning or related technical field
- Experience with foundation models (text, audio, or multimodal) and in-depth knowledge of the latest advancements in the multimodal domain
- Hands-on experience with running large-scale experiments
- Proficiency in Python and modern ML frameworks such as TensorFlow, PyTorch or JAX
- Excellent interpersonal skills and ability to work in a team, as well as independently
- Experience with multilingual data and understanding of the complexities and tradeoffs involved when scaling to non-English languages
- Hands-on experience with building and running large-scale data pipelines
- Publication record in relevant conferences demonstrating ability to conduct innovative research in deep learning or a track record in applying deep learning techniques to products