Predictive Maintenance for Manufacturing
Unplanned downtime is one of the most expensive things that can happen on a factory floor. The goal: spot it coming, and explain why, before the line stops.
Engineering record. Delivered by Jonathan Lam, Lamalo’s lead engineer, as a software developer and AI consultant at Lemey.AI, before the studio was founded. Clients are not named for confidentiality.
- Inputs
- Sensor + ops data
- Predicts
- Likely downtime
- Explains
- Root causes
- Delivery
- Automated CI/CD
The challenge
The plant already had data — machine sensors and operational records — but it was used after the fact, to explain what had gone wrong. The client wanted the opposite: warnings early enough to act on, and a clear pointer to the likely cause, so the maintenance team fixes the right thing first.
How it works
- CollectPlant data
Sensor readings combined with operational records.
- LearnPython ML
Models trained on the plant’s own history of readings and events.
- FlagEarly warning
Potential downtime detected before it happens.
- ExplainRoot cause
Likely causes surfaced to support the maintenance decision.
How it was done
- Machine-learning models in Python built on the plant’s own sensor and operational data.
- Diagnosis, not just alarms: the system points toward root causes, which is what turns a warning into a decision.
- Engineering hygiene — Git source control and automated CI/CD pipelines so models could be improved continuously without breaking what worked.
The outcome
Better plant up-time and better decision support: maintenance moving from reacting to anticipating.
Same year: an AI-readiness review for air defence
Not every AI project starts with a model. For a client in the air-defence sector, Jonathan led interviews, analysis and reporting on AI readiness and digital transformation — assessing infrastructure, data pipelines and software architecture, and recommending where AI could genuinely improve operations, how to modernise systems, and how to deploy securely in a high-stakes environment.
What this means for your business
It is the model for our AI discovery sessions today: start from the data you already have, find the one or two places AI will clearly pay off, and build it so it keeps working after launch. Book an AI discovery session →