Neuromorphic Inference Lab Neuromorphic Inference Lab

Hero System

NeuroGrid Fault Risk Scoring Platform

A production-first ML system that predicts medium-voltage fault risk to prioritise preventive interventions. Demonstrates the full delivery loop: ingestion → feature engineering → training (MLflow) → artefact versioning → Dockerised serving (FastAPI) → live inference.

Open evidence dossier Live API docs Source
CI/CD for ML Inference Serving Artefact Promotion Feature Engineering MLflow Tracking Monitoring-ready outputs

Problem → impact

Grid operations need a consistent way to rank assets/feeders by fault risk and allocate preventive interventions. A risk score supports earlier dispatch decisions and reduces exposure to unplanned outages.

Proof links

Recruiter-friendly verification paths (code + CI + release artefact + live endpoint).

Evidence dossier CI training runs Model artefact Proxy health
Time-aware split Reproducible builds Deployment-ready artefacts

Live inference demo

Score risk via a same-origin Cloudflare proxy (no CORS). If the upstream service was idle, the first request may take a moment to warm up.

Open Swagger

ATS keyword pack (explicit)

Full-Stack Machine Learning Engineer, Data Scientist, scalable ML pipelines, feature engineering, statistical modelling, MLflow experiment tracking, CI/CD for ML, Dockerization, FastAPI model serving, inference optimisation (patterns), artefact versioning, automated retraining (design pattern), monitoring-ready outputs, reproducible builds, time-based split.