About
I'm completing a Master's in Financial Engineering at WorldQuant University. My research so far covers portfolio construction under mandate constraints, asset allocation across volatility regimes, return forecasting with neural networks, and how model selection goes wrong.
I came to this from Agribusiness Management, by way of Software Engineering, Cloud Computing, Data Analytics and Business Analysis. Financial engineering is where those finally pointed in the same direction. I trained as an engineer before I trained as an analyst, which means I build the analysis rather than specify it — data through to something someone can read. I report what the data supports. Two of my four research projects returned findings that contradicted the original idea, and both are written up here, because a strategy that doesn't work is worth knowing before money goes near it.
Currently Software Engineer at Siscom Africa, building backend services and data features for fintech platforms, while finishing the MSc. Looking for analyst work spanning financial, investment, research or credit, in Nairobi or remote.
I started on the credit side, assessing whether borrowers could repay — income, existing obligations, repayment capacity — and writing the recommendation that went to the approval committee. That is where I learned that an analysis is only worth what the person reading it can do with it.
I then trained as a software engineer and spent two years building and shipping full-stack products, which taught me to work with data at scale rather than in spreadsheets. Those two threads met in financial engineering: the quantitative reasoning of credit and markets, executed with the tooling of an engineer.
I run the full workflow — source and clean the data, explore it for structure, select and test models, then explain what the results mean in practical terms. Nothing is skipped because it is tedious; most analytical errors are made before the model is ever fitted.
I report what the data supports, including when it disproves the idea. Two of the three research projects on this site returned a negative result, and both are here because a strategy that does not work is worth knowing before capital is committed to it.
What I do
Pipelines and distributed processing in Spark, PySpark, Hadoop and Hive when the dataset outgrows a single machine, with reporting in Power BI and Python.
Regression, classification and clustering through to neural networks, with the out-of-sample discipline that separates a real result from an overfitted one.
Return and volatility analysis, correlation and diversification structure, asset allocation testing, and backtesting strategies against honest benchmarks.
Currently
Software Engineer at Siscom Africa, building backend services and data features for fintech and AI platforms, while completing the MSc. Open to Analysts roles spanning Data, Financial Analysis, Investment Analysis, Research Analysis, or Credit Aalysis roles in Nairobi and to remote work.