Data Scientist (AI-Native) — Growth & Credit

About Umba

Umba is a pan-African digital bank with operations in Kenya and Nigeria. Its goal is to make financial services easier to access, more affordable, and more useful for people across Africa.

The company is reimagining banking through intelligent, automated financial products built with machine learning. Its platform supports digital banking, lending, and payments across Android, iOS, and web, and serves both individual and business customers at scale.

Based in Nairobi, Umba acquired a licensed deposit-taking microfinance bank in 2023 and has expanded revenue by more than six times since then. The company is seeking ambitious, high-energy people who believe technology can create financial freedom and help build Africa’s leading digital bank.

Role Overview

As the Data Scientist focused on growth and credit, you will own the two model areas that directly drive profitable expansion: customer acquisition and underwriting.

On the credit side, you will develop and keep improving scoring systems that determine who can be lent to and on what terms. These models will use data such as bank statements, payment history, CRB records, and behavioural signals collected within the app. The same scoring framework must work for both digitally acquired customers and those sourced by the sales team, so the solution needs to support both workflows.

On the growth side, you will improve how marketing spend is allocated to bring in customers. That includes ad targeting, funnel conversion, channel attribution, and experimentation to understand which actions improve CAC and LTV. You will manage the full loop from target selection to conversion to repayment.

This role differs from a standard data science position. Umba operates in an AI-native environment, where tools such as Claude Code, Codex, and other LLM-based systems are used to speed up analysis, generate model code, create pipelines, and iterate quickly. The role therefore places increasing emphasis on defining precise problem statements that AI systems can execute, reviewing and strengthening AI-generated work, building automated feedback loops, and setting the quality standard for production-ready models.

You will work closely with Engineering, Product, and Sales to launch models that influence lending decisions immediately. This is a highly technical, in-office position in Nairobi, and it requires strong ownership in a small, high-performing team.

Responsibilities

• Develop, deploy, and continuously refine credit scoring models using bank statement data, payment histories, CRB pulls, and in-app behaviour signals.
• Create automated underwriting flows for both digitally sourced customers and applications brought in by the sales team.
• Build retraining pipelines so scoring improves continuously as repayment outcomes come in, rather than on a quarterly cycle.
• Track model performance, detect drift, and set up automated alerts for issues in production.
• Work with Risk and Operations to define policy thresholds, override rules, and human review processes around model decisions.
• Improve ad targeting across acquisition channels, including audience selection, bid strategy, creative performance, and lookalike audience design.
• Measure the acquisition funnel end to end, from impression and click through install, KYC, first loan, and repayment.
• Design and run A/B tests for acquisition and product journeys, and build the testing infrastructure so the team can run experiments independently.
• Create attribution, LTV, and CAC models that can support real business decisions.
• Write detailed technical specifications that AI-assisted workflows can execute reliably.
• Use AI tools such as Claude Code, Codex, and similar systems to accelerate data cleaning, feature engineering, and analysis while carefully validating every output.
• Extend the data platform by integrating new sources such as third-party APIs, CRB providers, and payment rails when models require them.
• Clean, process, and verify data integrity, especially for anything affecting lending decisions.
• Communicate findings clearly to non-technical stakeholders and defend recommendations with evidence.

Requirements

• At least 4 years of hands-on experience in data science or applied machine learning in production settings.
• Strong command of Python, including pandas, scikit-learn, and numpy, plus SQL capability that allows you to move from raw data to deployed model without relying on engineering support.
• Practical experience with classification and regression problems, including feature engineering, model selection, calibration, and evaluation under class imbalance.
• Solid understanding of applied statistics, including hypothesis testing, regression, experimental design, and handling selection bias and censored outcomes.
• Experience working with messy financial datasets such as transaction records, bank statements, payment data, and credit bureau data, or strong evidence that you can ramp up quickly.
• Comfort working with relational databases such as Postgres or MySQL and with modern data tooling.
• Strong written and verbal communication skills, with the ability to explain model behaviour to credit officers, marketers, and engineers.
• Valid work authorisation for Kenya.
• Equal consideration is given to qualified candidates regardless of sex, gender identity, sexual orientation, race, colour, religion, national origin, disability, veteran status, age, or other protected characteristics.
• Candidates with disabilities or special needs can request accommodation.

Preferred Experience

• Background in credit scoring or fraud modelling, especially in emerging markets or thin-file populations.
• Experience in marketing analytics or growth experimentation, including work with Meta, Google, attribution, and funnel analysis.
• Production ML experience covering deployment, monitoring, and retraining pipelines.
• Regular use of AI coding tools such as Claude Code, Codex, or GitHub Copilot in day-to-day analysis and modelling work.
• Ability to write structured technical specs and prompts that consistently produce reliable code and analysis.
• Strong review skills for catching subtle issues in AI-generated SQL, features, and pipelines that may pass tests but still produce incorrect results.
• Understanding of common failure modes in AI-assisted analysis, including leakage, hallucinated joins, and plausible but incorrect feature definitions.

Bonus Experience

• Exposure to payments, lending, or fintech in Kenya or across Africa.
• Familiarity with CRB providers and Kenyan banking data formats, including Metropol, TransUnion, and CreditInfo.
• Experience deploying LLM-based features into production data products.

What We Value

• Data scientists who think in systems and feedback loops rather than one-off models.
• People who can use AI to ship much faster without sacrificing rigour.
• Builders who can own a problem end to end, from data to model to deployment to monitoring.
• Pragmatic operators who prefer shipping a working version now instead of waiting for a perfect solution later.

Work Status

This role requires valid work authorisation for Kenya and is based in Nairobi, Kenya, with an in-office work arrangement.

Equal Opportunity

Umba is an equal opportunity employer. All qualified applicants are considered without discrimination based on protected characteristics. Accommodation is available for candidates who need disability or special-needs support.