Our team and mission

Within the Earth Observations Projects Department, the Copernicus Sentinel Expansion missions are currently under development. One of those missions is the CHIME (Copernicus Hyperspectral Imaging Mission for the Environment). CHIME plans to carry a unique visible to shortwave infrared spectrometer to provide routine hyperspectral observations to support new and enhanced services for sustainable agricultural and biodiversity management, as well as soil property characterisation.

You will be able to join the CHIME project, participating as part of the system engineering and satellite engineering section in activities related to Onboard Satellite Artificial Intelligence Processing and the evaluation of end-to-end mission performances.

Further details on the CHIME mission can be found at:

• ESA - Copernicus Sentinel Expansion missions 
• ESA - Going hyperspectral for CHIME 
• Copernicus Hyperspectral Imaging Mission for the Environment (CHIME): an overview of its mission, system and planning status; SPIE Proceedings Volume 12729; 1272909 (2023) The copernicus hyperspectral imaging mission for the environment (CHIME): an overview of its mission, system and planning status 
• Onboard Processing of Hyperspectral Imagery: Deep Learning Advancements, Methodologies, Challenges, and Emerging Trends Onboard Processing of Hyperspectral Imagery: Deep Learning Advancements, Methodologies, Challenges, and Emerging Trends | IEEE Journals & Magazine | IEEE Xplore 
• CHIME's end to end simulator (CHEES): simulating, testing and generating hyperspectral products CHIME's end to end simulator (CHEES): simulating, testing and generating hyperspectral products | SPIE Sensors + Imaging 

Candidates interested are encouraged to visit the ESA website: http://www.esa.int

Field(s) of activity for the internship

Topic of the internship: CHIME: Foundation Models & On-Board Intelligence for Hyperspectral Earth Observation

Join us to advance artificial intelligence for the Copernicus hyperspectral imaging mission for the environment. You will design methods that combine large “foundation” models with ultra-light detection running on the satellite to deliver trustworthy, low-latency environmental insights for round-the-clock monitoring and early warnings (for example, marine spills, algal blooms, or fires).

What you will do

• Build and adapt foundation models. Pretrain and refine models on spectral–spatial image cubes from simulators and existing satellites, then tailor them to environmental anomaly detection;
• Distil models for flight constraints. Translate foundation-model features into a compact linear support vector machine that runs in real time on power-limited hardware;
• Validate with an end-to-end mission simulator. Use an existing processing chain (from raw sensor data to geophysical products) to benchmark accuracy, latency, reduction in downlink volume, and robustness across scenes, illumination, and noise;
• Increase transparency and trust. Add explainable methods (for example, identify which wavelengths drove a decision) and calibrated uncertainty so each alert carries an interpretable rationale and confidence;
• Reduce false alarms. Calibrate decision thresholds, mine difficult examples, and apply simple post-processing (for example, morphological filters) to meet strict limits on false alarms per orbit while preserving detection quality;
• Explore continuous monitoring. Prototype night-time hyperspectral use cases and selective downlink policies that prioritise the most informative image excerpts.

Learning objectives

By the end of the internship, you will be able to:

• Pretrain, design and fine-tune a foundation model for hyperspectral imagery and evaluate transfer to multiple environmental tasks;
• Distil complex models into efficient, transparent decision functions suitable for execution on board a satellite;
• Run end-to-end simulation campaigns and report mission-relevant metrics (for example, detection quality, latency, and savings in downlink volume);
• Produce interpretable outputs with explainable methods and calibrated uncertainty, and design procedures that reduce false alarms;
• Communicate results through clear reports, open and reproducible code, and—where appropriate—publish or present your work at conferences and workshops.

How we work

You will collaborate with scientists and engineers from the mission and from downstream user communities. 

We value open, well-documented code and reproducible experiments. You will have access to the mission’s end-to-end simulator, curated datasets, and mentoring on both scientific and engineering topics.

If you are motivated by the idea of turning cutting-edge machine learning into operational environmental monitoring from space—and you enjoy balancing scientific depth with practical constraints—this internship is for you.

Behavioural competencies

Result Orientation
Operational Efficiency
Fostering Cooperation
Relationship Management
Continuous Improvement
Forward Thinking

For more information, please refer to ESA Core Behavioural Competencies guidebook

Education

You must be a university student, preferably studying at master’s level. In addition, you must be able to prove that you will be enrolled at your University for the entire duration of the internship.

Additional requirements

The working languages of the Agency are English and French. A good knowledge of one of these is required. Knowledge of another ESA Member State language is an asset.

During the interview, your motivation for applying to this role will be explored.

You should also have:

• Strong maths/signal-processing base;
• Python proficiency;
• Experience with simulators (digital-twins) and spectral or image data.

It would be an advantage to have:

• Familiarity with remote sensing basics, explainable methods, and uncertainty;
• Deep-learning libraries, quantisation/pruning, or compact models.