Senior Data Scientist, MNO Data

About

GiveDirectly is hiring a Senior Data Scientist to improve, adapt and make repeatable its methods for residence-location inference from Mobile Network Operator data. The role works with operator engineers on technical specifications and implementation, supports privacy and security work, and involves substantial international travel; Kenya is among the countries where GiveDirectly can employ candidates. Salary: Kenya Base Salary: $72,550. Salary: Bonus at Target Performance: 10%. Salary: Annual Benefits Stipend: $8,454. Duty station: Remote. Terms of service: Substantial international travel, roughly 25 to 50% of your time depending on emergencies, sometimes at short notice. Terms of service: Reports to: Senior Product Manager - MNO data. Job grade: Level: Senior Manager.

How to apply

Apply online through the official advert page on Greenhouse. Submit a Resume/CV and complete the required application questions, including your current location, visa-sponsorship status, and two technical experience questions.

Responsibilities

• Improve and extend our library of analytical method templates for establishing where people live from MNO data, and depending on circumstances, for detecting displacement or estimating poverty.
• Design novel methods when an intervention requires it.
• Support colleagues in technical conversations with operators, bringing clarity on the data we need and why, and steering discussions toward technical designs that make the analysis workable.
• Assess the information that can be inferred from each operator's data given data types and sparsity, and for each geography and targeted area.
• Support intervention design to account for data quality.
• Write tests and method specifications for MNO data engineers to implement: steps, parameters, pseudocode, validation criteria.
• Guide operator engineering teams through implementation, in writing and in person.
• Adapt methods to accommodate large data volume and computational constraints, balancing accuracy and operational needs.
• Run in-country workshops with operators to support the set-up of new analyses and - where feasible - tune parameters based on statistical tests and distributions.
• Feed every new analytical and implementation element back into standard templates, so the next country or intervention starts from a better position than the last.
• Design for privacy from the start, so that as little data as possible has to be shared: aggregation, minimization, and where useful, privacy-enhancing technologies such as private set intersection or homomorphic encryption.
• Protect the confidentiality of what operators share with us and improve internal data security processes and tools as we grow.
• Build internal training material so that our technical and non-technical teams understand the strengths and limits of this data, and design interventions accordingly.

Qualifications

• Exceptional alignment with GiveDirectly Values and active demonstration of our core competencies: emotional intelligence, problem solving, project management, follow-through, and fostering inclusivity.
• Strong mathematical modeling, machine learning and statistics.
• Deep comfort with spatio-temporal data: temporal filtering, GIS, probability distributions, combinatorial reasoning, and working with sparse, irregular observations.
• The ability to reason without ground truth.
• Pragmatism about the trade-off between accuracy and implementability.
• Excellent communication and relationship-building skills.
• Experience writing technical specifications that other teams implemented, clear enough that people you don't manage could follow them correctly.
• Able to read and review SQL and Python written by others, and to find the logical error in someone else's query.
• Enough understanding of large data volumes and databases to judge when analytical simplification is required.
• Commitment to privacy by design as a foundational value and starting point rather than a compliance step: designing analyses so that less personal data has to be shared.
• Commitment to confidentiality.
• Experience designing training material that changed how non-specialist colleagues used a data source.
• Comfort in fast-moving contexts where the brief may often shift, and where your technical expertise and understanding may not be shared by other stakeholders.
• Comfortable with travel to fragile and post-disaster contexts.
• Language requirement: English.