Home
Hello, I'm Dan Kibiwott, a full-stack software developer and machine learning engineer based in Cambridge, United Kingdom. My recent work spans childcare workforce platforms, coaching software, finance tools, and Python research systems.
I work across product interfaces, APIs, databases, payments, and deployment, with a particular interest in turning complex workflows into practical software.
Current Tech Stack
Web applications
TypeScript · Next.js · React · Tailwind CSS · Radix UI
APIs & data
Node.js · Express · GraphQL Yoga · Prisma · PostgreSQL
Product & operations
Auth.js · Stripe Connect · Resend · Sentry · Docker · Caddy
Research & automation
Python · scikit-learn · MetaTrader5 · MQL5 · Windows OCR
Featured Projects
| Name | Description | Links |
|---|---|---|
| CKC Induction Platform | A staff induction and operations platform for Cambridge Kids Club, with dedicated staff, playleader, and admin workspaces. It brings together learning modules, evidence submissions, task reviews, practical training schedules, staff availability, and progress reporting. #Next.js 15#React 19#TypeScript#Express 5#Prisma#PostgreSQL#TanStack Query#Editor.js#Google Cloud Storage#Docker#Caddy | Details |
| Childcare Workforce Systems | A website for presenting childcare workforce software, with case studies for staff induction and prototypes for recruitment and staff administration. #Next.js 16#React 19#TypeScript#Tailwind CSS#Lucide | Website |
| Staff Induction Platform Showcase | An interactive frontend showcase of staff, manager, and admin workflows, using fictional people and records. A guided walkthrough follows tutorial publishing, practical training requests and sign-off, task feedback and resubmission, supervision, and progress exports. #JavaScript#HTML#CSS#localStorage#Node.js#Playwright | Details |
| Lab 004 - Proprietary Capital Systems | A months-long machine learning research and execution system that extended the ideas behind Arasaka into a complete experimental capital pipeline. I eventually stopped the live experiment after the evidence did not justify continuing it, but the project remains one of my strongest research and systems-engineering case studies. #Python#MetaTrader5#scikit-learn#MQL5 | Article |
| Gym Primer | A multi-role coaching platform for personal trainers, clients, and platform admins. It supports coach workspaces for client management, workout planning, operations, packages, billing, and payments. #Next.js 16#React 19#TypeScript#Prisma#PostgreSQL#Auth.js#Stripe#Stripe Connect#Resend#Sentry | Website |
| Finance Dashboard | A personal finance dashboard with statement imports, bill analysis, paid-bill checklists and uploads, saved budget plans, monthly rollover, and visual reporting. #React 19#TypeScript#Vite#Tailwind CSS 4#GraphQL Yoga#PostgreSQL#Recharts#Render | Details |
Featured Articles
Lab 004: Proprietary Capital Systems
Lab 004 was a months-long attempt to turn a forex ML idea into a complete research system: data ingestion, features, regimes, model routing, strategy rules, backtests, risk reports, MT5 execution, scheduled jobs, logs, and guard rails. It did not become the performant trading system I hoped for, but it became a serious engineering and research artifact.
Building Arasaka Neural Bastion
Arasaka Neural Bastion is my experimental forex machine learning research pipeline. This article walks through the architecture, model loop, deployment thinking, and why the most useful part of the project is not predicting the market, but building a system that can reject weak assumptions quickly.
SQL Joins Explained
One of the most powerful tools when working with relational data in SQL is Join Queries. SQL allows us to combine data from multiple tables into a single result set which enables us to retrieve data spread in multiple tables just using a single command. In this article, we'll explore the different types of SQL joins, focusing on how they work in PostgreSQL.
Building a Chess Web App
In this article, we will be building a Chess application using React that will allow the player to drag pieces across the board. We'll be using TailwindCSS for styling. To view what we will be building by the end of this article, visit dankibiwott.com/experiments/0001.
Machine Learning for Imaging
An evaluation of classical machine learning, feature extraction, and deep learning methods for binary image classification (Raccoon vs Rifle) using a curated OpenImages dataset.