I'm a final-year Computer Science student at Strathmore University in Nairobi, Kenya, building software where machine learning, cloud infrastructure, and financial markets meet. I like taking systems from proof-of-concept to production — whether that's an MLOps pipeline for a trading agent, a tax-integration app for the Kenya Revenue Authority, or a game with a real payment system attached. Off the keyboard, I trade gold (XAU/USD) using ICT and Smart Money Concepts, which usually finds its way back into the bots and indicators I build.
- 🔭 Current Focus: A Deep Reinforcement Learning trading agent and its surrounding MLOps pipeline for hydroelectric energy markets, built as my final-year capstone — alongside Krekit, a physics-based pool game with a live wallet system.
- 🌱 Learning & Exploring: Distributed systems consensus algorithms, NewSQL/NoSQL data stores, and deeper Deep RL techniques for trading environments.
- ⚡ Core Interests: ML Engineering, FinTech & Algorithmic Trading, Cloud-native MLOps/DataOps, Distributed Systems, and Game Development.
- 📌 Currently: Interviewing for Software/ML Engineering roles, including an internship with ishango.ai.
Capstone Project — Python · AWS · Terraform · Airflow · Docker · MLflow
End-to-end automated DataOps/MLOps pipeline for an energy-trading domain (hydroelectric power). Infrastructure is provisioned with Terraform across AWS S3, Glue, SageMaker, App Runner, and EC2, mocked locally with LocalStack and Docker, and orchestrated with Apache Airflow. A Deep Reinforcement Learning (DRL) agent is trained on engineered technical-indicator features and tracked through MLflow.
Godot Engine · Flask · MongoDB · M-Pesa
A web-deployed pool game optimized with Brotli/Gzip compression and a custom Flask server configuration to resolve engine decoding errors, with browser-compatibility checks and a landscape-mode lock for mobile play. Backed by a MongoDB wallet system with M-Pesa STK Push deposits and a custom secure-ticket system that protects player funds if the game crashes.
Professional Attachment @ Oraram Investments — React Native · Flask · MongoDB
An app automating eTIMS tax invoicing for traders and petrol stations by integrating directly with the Kenya Revenue Authority. Built the initial proof-of-concept, planned the scalable system architecture, and authored the API documentation.
MQL5 · Pine Script — MetaTrader 5 & TradingView
Expert Advisors with defensive handling of overnight spread widening, time filters, spread controls, and dynamic trailing stops. Pine Script indicators automate Smart Money Concepts/ICT setups — QM entries, liquidity sweeps, SMT divergence — alongside multi-timeframe support/resistance detection and MACD/RSI/200-EMA hybrid trend filters, with "New Bar" checks to prevent signal repainting.
Flask · MongoDB
A shared vision-board app for collaborative goal-setting, a Secret Santa app for holiday gift coordination, and a front-end car-themed website.
- ML Model Diagnosis & Evaluation — Diagnosed Logistic Regression and kNN classifiers across accuracy, F1-score, generalization, and data-leakage detection.
- Forest Sound Classification (Concept) — Designed a classification approach to detect deforestation activity from forest audio recordings and alert environmental authorities.
- Distributed Databases & Consensus — Presented on global consistency, consensus algorithms, replication, and fault tolerance in modern NewSQL/NoSQL systems.
- 📧 Direct Contact: cglynn.skip@gmail.com