Hardship Level (not applicable for home-based)H (no hardship)
Family Type (not applicable for home-based)
Family
Staff Member / Affiliate TypeCONS International
Target Start Date2025-02-03
Deadline for ApplicationsDecember 30, 2024
Terms of ReferenceAs per the Terms of Reference. Recruitment of a consultant as Associate Data Scientist (AI expert - at PhD level) for Innovation Project: Early Warning and Effective Response System. To be based at the Luxembourg Institute of Science and Technology in Luxembourg.
Title of project: Early Warning and Effective Response System (EWERS).
Purpose of project (summary): The project aims to enable UNHCR and partners in country operations to 1) improve their capacity to forecast humanitarian emergencies and the likelihood/impacts of displacement through country and context-specific models, 2) enhance informed investment planning for resilience-strengthening efforts with local communities, 3) make time-sensitive and data-informed decisions to prepare for and respond to emergencies through strengthened multi-stakeholder cooperation.
General Background of Project or Assignment, Operational Context:
In line with the 2030 Agenda for Sustainable Development and the principle of “Leave No One Behind,” UNHCR aspires to develop, together with strategic partners, a people-centred global Early Warning and Effective Response System (EWERS) that will systematically collect timely and effective data and information on potential risks and threats related to forced displacement caused by conflict and/or natural hazards. The EWERS will provide credible and actionable information and analysis at high frequency and high geographical resolution, which could:
• enable UNHCR and partners in country operations to improve their capacity to predict humanitarian crises and the likelihood/impacts of displacement through a country and context specific tool.
• ensure UNHCR and the partners/stakeholders with whom UNHCR shares the alerts will be able to conduct informed investment planning for strengthening resiliency efforts within local communities, as well as make time-sensitive and data-informed decisions to prepare for and respond to crises through strengthened multi-stakeholder international cooperation; and
• allow UNHCR and partners to harness emerging technologies and tools, such as nowcasting, forecasting and machine learning-based models, for anticipatory humanitarian action.
The project is led by the Division of Emergency, Security and Supply, in coordination with Innovation Service and Global Data Service. The project is implemented in coordination with the Luxembourg Institute of Science and Technology (LIST).
Purpose and Scope of Assignment:
The Associate Data Scientist will report to the Senior Project Manager and will work closely with UNHCR and LIST colleagues as well as external parties. S/he will work in the premises of the LIST and embedded in the LIST remote sensing and natural resources modelling group to work on the project. By the end of 2025, the project team including the Associate Data Scientist is expected to develop a minimum viable product which provides displacement forecast over early warning system(s) for selected pilot areas. The team will engage in the user requirement analysis and data collection starting January, will design and develop displacement forecast models starting March, will build a visualization platform starting April, and will pilot and evaluate the system starting September to reach the minimal viable product goal. The Associate Data Scientist is a core member of the project team and will assume following responsibilities in coordination with UNHCR - LIST project team:
Research and preparation:
• Conduct research to analyze externally available early warning models, based on Uppsala University’s external scoping and other resources, to build displacement models on.
• Conduct research in artificial intelligence and explore algorithms and methodologies.
• Understand displacement risks, potential numbers, and geographical scopes in relation to triggering events.
• Determine displacement triggering early warning events to pursue in different phases of the project.
• Identify pilot countries and focus area.
• Identify data requirement and their availability.
• Identify open data sources.
• Establish data requirements.
• Analyze potential risks in relation to data and develop mitigation strategies.
Design and development:
• Develop multimodal deep learning models to ingest multi-source data, including social media data, climate data, Earth Observation (EO) data, UNHCR’s registration date etc.
• Design multivariate time series based early warning system by exploring modern deep learning models, such as LSTM, GNN, and Transformer.
• Develop uncertainty quantification methods to calibrate and estimate uncertainty of forecast.
• Develop (deep) causal inference/discovery algorithms to uncover causes and drivers of early warning.
Data collection:
• Develop processes to extract, clean, and analyze datasets.
• Collect, clean, store, and organize data in line with UNHCR’s Data Management Guidelines.
• Create and manage a directory in UNHCR designated cloud services to safeguard the data collected.
• Categorize, update, track and analyze data.
• Create data blocks to be automated and included in future applications.
Pilot, evaluation and rollout:
• Model evaluation and debugging.
• Analyze and incorporate end user feedback in relation to the displacement forecasting and user interface application.
• Pursue deployment of the system.
Documentation:
• Document meta data.
• Document sources, resources, development process, results (including errors), and decisions.
• Provide inputs to communication materials, including scientific articles.
Required qualifications, language(s) and work experience:
PhD in a field related to machine learning, artificial intelligence, human interaction systems, remote sensing, image or signal processing, applied mathematics, computer engineering, telecommunications engineering, or computer sciences (or similar).
Field of expertise, competencies:
Required/mandatory:
- Experience in data mining.
- Experience in developing time-series forecasting systems.
- Experience in modern deep learning models, such as LSTM, GNN, and Transformer.
- Excellent programming skills (e.g., Python, C/C++, etc.).
- Knowledge on forced population displacement.
- Knowledge on database technologies (e.g. CouchDB, SQL)
- Solid understanding of machine learning/deep learning methods, statistical modeling, and optimization techniques.
- Skills in presenting scientific research, writing papers in scientific journals, and crafting technical reports.
- Communicative and willing to learn, self-organized, and creative.
- Ability to work both independently and collaboratively in an international team.
Desirable:
- Experience in Earth Observation (EO) data processing; experience with data visualization technologies (e.g. Tableau, RShiny, MarkdownR, MS Vision, MS InDesign/Visual Studio); familiarity with commercial artificial intelligence and machine-learning software (e.g. Watson, Crimson Hexagon, Eureqa, etc); experience with cloud computing services.
,
,
,
,
Skills
EducationDoctor of Philosophy (PhD) (Required)
Certifications
Work Experience
Other informationThe hiring location is Geneva, Switzerland. The consultancy is based in Luxembourg. The consultant will be working on the LIST premises with the remote sensing and natural resources modeling group.This position doesn't require a functional clearance
Home-BasedNo