I design reliable data pipelines, real-time transaction streaming systems, and lakehouse platforms, with a focus on East African fintech infrastructure.
- 🏗️ I build batch and streaming pipelines for transaction data: ingestion, transformation, quality checks, and observability
- 💳 Deep interest in M-Pesa / Daraja data: reconciliation, fraud signals, and careful transaction-state handling
- 🧱 I care about systems that are dependable, observable, and easy to hand over, built so the next person can understand them
|
💸 Fintech & mobile money 🌍 Technology for Africa 🧠 Machine learning & AI 🚀 Building products from scratch |
✍️ Writing & sharing what I learn 🤝 Open source & collaboration 📚 Continuous learning 🎬 Film & tech culture |
flowchart LR
A[Sources<br/>M-Pesa · APIs · DBs] --> B[Kafka<br/>event streams]
B --> C[Flink / Spark<br/>stream + batch]
C --> D[(Bronze → Silver → Gold<br/>lakehouse)]
D --> E[dbt<br/>models + tests]
E --> F[Dashboards & APIs<br/>Streamlit · Grafana]
G[Airflow<br/>orchestration] -.-> C
G -.-> E
| Area | What it looks like in practice |
|---|---|
| ETL / ELT | Reliable workflows for structured and semi-structured data, with batch loading and validation |
| Real-time streaming | Event-driven ingestion and processing with Apache Kafka, Flink, and Spark |
| Modeling & transformation | Dimensional models, incremental dbt transformations, and data-quality checks |
| Orchestration | Airflow DAG design, dependency management, retries, and backfills |
| Financial data | Reconciliation-oriented workflows and fraud/anomaly signals on transaction data |
| Storage & lakehouse | PostgreSQL, Cassandra, Redis, and Delta Lake |
| Reliability | Pipeline metrics, logs, alerts, and data-freshness checks |
| Platform & DevOps | Containerized environments, Terraform, Kubernetes, and cloud deployment on AWS |
Languages
|
Streaming & Orchestration
|
Databases & Storage
|
Analytics & Observability
|
Infrastructure & DevOps
|
Data Quality
|
Real-time M-Pesa transaction streaming pipeline: ingest, process, stream, and analyze events end to end.Kafka · Spark · PostgreSQL · Docker
|
Streaming fraud engine with ML risk scoring and explainable signals (velocity, SIM-swap correlation, mule networks) feeding a decision engine.Kafka · Flink · dbt · PyTorch
|
Fault-tolerant batch pipeline that cleans, validates, and loads 1.6M+ records with automated data checks.Python · PostgreSQL · Docker · Terraform · AWS
|
Ingests public economic indicators, models them with dbt, and serves query and CSV APIs with signed cursor pagination, Redis ETag caching, and rate limiting.dbt · PostgreSQL · Redis · Python
|
Real-time transaction monitoring with rule-based anomaly scoring, runnable locally or on AWS.Python · S3 · Kinesis
|
ETL, streaming, fraud detection, and financial analytics across Kenyan fintech and banking data, with live dashboards.Python · Spark · Streamlit · Grafana
|
Have a pipeline problem, a collaboration idea, or an opportunity? Reach out.