[Hiring] Machine Learning Engineer @Kora

Role Description

Kora is the marketplace for everything payments. We offer a robust payment API for payment collections, disbursements, and conversions for businesses anywhere in Africa. Our vision, which is at the core of what we do every day, is to create a world void of digital financial barriers. We are committed to delivering reliable, secure, and easy-to-use digital financial solutions to every single customer with a guarantee that it is improving their lives. To achieve this mission, we need people like you.

We run payments across Africa and are now positioned as a global fiat and stablecoin payment infrastructure. We offer mobile money, virtual bank accounts, and virtual cards for payins and payouts across multiple markets. Our data infrastructure is batch-first (Airflow + a cloud data warehouse) and we use Vertex AI for our MLOps lifecycle. The ML team is high-ownership: you will build models, design systems, ship them, and observe them in production.

You will work on merchant-facing intelligence: forecasting, anomaly detection, segmentation, as well as automation and product-layer ML. If you want to build practical things that matter in a context that most ML engineers never get near, this is the role.

What You'll Work On

• Design and ship a per-merchant payment volume forecasting system: time-series decomposition, Africa-specific event calendars (salary cycles, MNO maintenance windows, public holidays), quantile regression for uncertainty bounds.
• Build and maintain fraud/anomaly detection across the payment stack (residual-based and model-driven) with tiered alerting logic mapped to merchant risk profiles.
• Own the dynamic merchant segmentation system end-to-end: rule-based and data-driven hybrid, percentile thresholds grounded in EDA, segment-transition features as ML inputs.
• Instrument and monitor deployed models: drift detection, retraining triggers, and evaluation pipelines via Vertex AI.
• Build automation tooling that sits alongside the core ML work: Airflow DAGs, pipeline scaffolding, and tooling to reduce operational toil.
• Contribute to product and strategic thinking.

Our Stack

• Apache Spark and Airflow
• Google Vertex AI
• Python
• SQL
• GCS/BigQuery

Qualifications

• 3+ years as an ML engineer in a production environment.
• Strong Python and comfort with Spark for large-scale data processing.
• Experience with time-series modelling: decomposition, forecasting, anomaly detection.
• Solid grasp of the ML lifecycle as a unified discipline.
• Ability to work with batch infrastructure and design for it deliberately.
• High ownership mentality: you notice problems and fix them as opposed to waiting to be assigned.
• Ability to identify gaps in data-driven business processes and come up with solutions.

Strong Plus

• Familiarity with Vertex AI (custom training jobs, model registry, pipelines, monitoring).
• Experience in payments, fintech, or any domain where label quality, distribution shift, and operational constraints are real problems.
• Exposure to African market dynamics.
• n8n or similar automation/workflow tooling experience.

Important

You will be evaluated less on credentials or certifications and more on the quality of your thinking. In this team, a strong ML engineer:

• Can explain why a design decision was made and what it trades off.
• Writes systems that the next person can understand and build on.
• Is honest about model limitations, especially in production contexts where overconfidence causes real loss.
• Closes the loop between model outputs and business outcomes without needing to be told to.

Benefits

• Health insurance.
• Sponsored and tailored training.
• Paid parental leave.
• Paid time-off.
• Flexible work style.
• Low-interest loans.
• Group Life Insurance.
• Access to up to four therapy sessions monthly.
• Day off on your birthday 🎂 🎁 🎉.
• Employee interest groups that provide supportive communities within Kora.
• Great company culture and the opportunity to work with a highly collaborative team building something great!

Note: We recognise imposter syndrome is real - any candidate who does not perfectly fit every characteristic of this role is still strongly encouraged to apply.