About the role
We are looking for a Research Engineer to help build end-to-end intelligence for spatial data and ML development. Someone who thinks in mathematics and builds in code, equally at home reading a frontier research paper and translating it into production software.
About Hyades
Hyades is a Spatial AI research lab on a mission to make machines understand the physical world. We are building the foundational intelligence layer for spatial data and machine learning that reasons natively across the full spectral and spatial domain, and the MLOps platform that puts that intelligence directly in the hands of enterprise geospatial ML teams.
We believe spatial data is the most underutilised source of signal on the planet, and that understanding it requires building new foundations rather than adapting existing ones. Our models are trained to reason about the physical world at the level of the data itself: spectrally, spatially, and temporally. Our platform, BlazerOps, deploys that intelligence into real ML workflows across insurance, climate, agriculture, and government.
What you will do
Research and Experimentation
- Design and implement rigorous training regimens that ground model behaviour in well-defined hypotheses.
- Monitor and evaluate automated agent-driven experimentation pipelines, validating outcomes against expected behaviour.
- Form clear hypotheses when models underperform and navigate agent workflows toward the correct solution.
- Maintain a deep and current understanding of ML research, especially in foundation models, spectral reasoning, and multimodal learning.
- Contribute to a research agenda focused on spatial data understanding across complex sensor modalities.
- Experiment with state-of-the-art techniques in model architecture, pre-training, and fine-tuning.
Communication
- Document experiments clearly and contribute to internal and external technical writing.
- Communicate findings across research and engineering teams with precision.
Requirements
- Master's degree or PhD in Computer Science, Physics, Mathematics, or a related field, or equivalent personal project experience.
- Strong mathematical foundation: understand key mathematical concepts that shape assumptions in machine learning and wider implementation systems.
- Familiarity with information theory and measure theory as tools for thinking about model learning and generalisation.
- Experience with foundation model pre-training and fine-tuning, in the geospatial domain.
Nice to have
- Experience with geospatial data: hyperspectral, multispectral, SAR, or LiDAR.
- Experience building agentic tools for automation.
- Experience training models on high-dimensional and time-series sensor data.
- Publications or open research contributions in ML or a related field.
- Proficiency with distributed training frameworks: Ray, Anyscale, or similar.
- Familiarity with cloud platforms: AWS or GCP.
- Physics intuition about sensor data and signal processing.