Industrial Data and AI

Practical application of data and AI to real industrial problems: process optimization, predictive maintenance on historian data, on-premise technical chatbots based on RAG, MLOps in regulated environments. Pragmatic approach, no experimental demos: everything we deploy must work in operation and survive a quality audit.

Our Expertise

Business Intelligence and operational dashboards (Power BI)

On-premise technical chatbots based on RAG for troubleshooting support

Process optimization through machine learning (Python, TensorFlow, Azure ML)

Predictive maintenance on historian data (PI System, Wonderware Historian)

MLOps and operational reliability of production models

AI validation in GMP environments (audit trail, explainability, traceability)

Internal team training for data tool appropriation

Industrialisation of ad-hoc analyses (critical Excel, scripts) into maintainable, versioned and auditable data tools

Technologies

PythonTensorFlowAzure MLPower BIRAG / LLM on-premiseMLOpsHistorianPredictive maintenance

Use Cases & References

On-premise maintenance chatbot based on RAG

Complete troubleshooting support solution for a pharma site under strict confidentiality constraints (cloud LLMs not allowed). Phase 1: Business Intelligence on failure history (10+ years of CMMS tickets) to identify recurring patterns. Phase 2: on-premise intelligent chatbot based on Retrieval-Augmented Generation consuming technical documentation, maintenance procedures and failure history to provide technicians with real-time contextual recommendations.

Technologies used:

Power BIRAG (Retrieval-Augmented Generation)LLM on-premiseVector database