I’m Chinonso Nwankwo, an ML engineer working on jet engines, oil pipelines, card payments and security robots. I take each problem from raw data to a live app, and I build the algorithms underneath from first principles.
Each one goes from raw data to an app a non-specialist can use. Open a case study for the decisions, the trade-offs and the results.
How long until this engine fails? Remaining-useful-life prediction on NASA C-MAPSS, framed as survival analysis so every estimate ships with a calibrated 90% interval a maintenance planner can trust.
Paste a free-text pipeline incident narrative and get back a structured record: what failed, what was done, what resulted, why it happened and how severe it was. A retrieval layer answers questions across the whole incident corpus.
Two logistic regression models written from scratch in NumPy score 283,726 card transactions, 0.17% of them fraud. Move the threshold and set the cost of a review to see exactly what each choice catches, misses and costs.
An autonomous security robot that maps a site, patrols on a schedule, detects people with YOLO and decides when to escalate. I founded Stratos AI & Robotics to take it from simulation to Nigerian banks and facilities.
Linear regression to PPO in NumPy, with no library doing the learning. Nothing is trusted until it passes four checks.
Synthetic data where the correct output is known in advance.
Finite differences on every hand-derived gradient, down to 1e-10.
Same inputs through scikit-learn or Stable-Baselines3.
A different algorithm for the same answer, like value iteration.
My foundation is in Computer Science at Enugu State University of Science and Technology.
I learn machine learning the slow way. Before I trust a model, I write it myself, check its gradients and make it agree with a reference implementation. That habit carries into everything I ship.
I aim it at industries where a wrong prediction has a real cost: aviation, oil and gas, finance and security. Rare events, imbalanced data and thresholds someone has to defend are the problems I enjoy most.
Foundations first, then applied systems, aimed at ML engineering where models touch physical assets.
Enugu State University of Science and Technology.
Supervised, unsupervised and reinforcement learning.
Engine RUL, pipeline NLP and fraud scoring apps.
Patroller complete in simulation. Hardware next.
Remote ML roles, plus AWS certification.
Aviation, energy, finance, security or manufacturing.
