Data Scientist

At Fido, we are building the future of finance in Africa, powered by advanced technology, data driven decision making and bold thinking.

Through AI, Cutting-edge data science and automation, we’re redefining how people access and experience financial services. Our goal is to make finance simple, smart and accessible, giving everyone the confidence to take charge of their financial story.

Joining Fido is an opportunity to drive real impact, solve meaningful problems and contribute to building a future where millions have the tools to create, grow and thrive.

‍

• Own the Model Lifecycle: Lead end-to-end credit risk model development from initial problem definition to production deployment and monitoring.

• Orchestrate Agentic Systems: Build autonomous agent networks using Claude Code and MCPs to automate feature engineering and model building.

• Develop Hybrid Pipelines: Combine tabular GBDTs (CatBoost, LightGBM) with LLMs to extract risk signals from unstructured data like graphs and time series data

• Enforce Evaluation Discipline: Create robust evaluation suites to continuously track model calibration, distribution shifts, and demographic fairness.

• Collaborate Cross-Functionally: Partner with Risk, Product, and Engineering to turn model insights into scalable credit limits and pricing strategies.

• Experience: 3 - 5+ years shipping high-impact production grade predictive models and decision systems.

• Education: MSc in a quantitative field (Computer Science \ Engineering \ Mathematics \ Statistics or similar) with focus on applied statistics, AI, machine learning, or related fields.

• Applied AI Capabilities: Experience deploying tool-using AI workflows and LLM APIs with schema-constrained, structured outputs.

• Engineering Orientation: Strong builder mindset focused on leveraging AI agents to deliver clean, production-ready pipelines with built-in testing and governance boundaries.

• AI Governance & Judgment: Strong analytical judgment to audit AI-generated pipelines, manage model risk, and enforce "human-in-the-loop" safety boundaries.