Client: one of the world's largest electronics manufacturers (name withheld under confidentiality). Location: on-site, Noida / Delhi NCR, India. Format: 4-day hands-on, concept-first with live coding on real production data. Audience: production (PE), quality (QA), and process engineers, plus plant IT — largely new to AI, working daily with MES/ERP data, OEE, and inspection lines.
What Was Covered
Four days — from demystifying AI to predictive analytics and computer vision
Day 1 · Demystifying AI for the Factory Floor
What AI is and is not in a manufacturing context; ML vs deep learning vs GenAI vs agentic AI (with factory examples); automation vs IoT; RAG, RPA and APIs; Industry 4.0 and 5.0; the analytics maturity model — descriptive → diagnostic → predictive → prescriptive; an honest AI-maturity assessment of the facility.
Day 2 · The Data Journey & AI-Assisted Analysis
The current workflow (MES/ERP → manual Excel → email) and its failure points; what makes data AI-ready; cleaning with Pandas and OpenRefine. Hands-on in Google Colab: load a production dataset, clean and visualise it in ~10 lines of Pandas, and auto-explain patterns with a GenAI tool.
Day 3 · Machine Behaviour, OEE & Automated Reporting
Reading machine behaviour from raw signals; calculating OEE, MTBF and MTTR in Python; building a live OEE dashboard in Apache Superset; GenAI-generated narrative shift summaries; querying multiple data sources in plain language with LangChain + a local LLM.
Day 4 · Pattern Detection, Predictive/Prescriptive & Computer Vision
Temporal, spatial, and correlational patterns; clustering (K-Means) and anomaly detection with Scikit-learn; predictive maintenance and RUL; prescriptive recommendations; the AI tool-selection framework; and computer vision & OCR — AOI, solder-paste inspection, component verification (YOLO, SAM).
Stack & Tools
Google Colab & Jupyter · Pandas · OpenRefine · Apache Superset · Scikit-learn (K-Means, anomaly detection) · Ollama with Llama 3 · LangChain · GenAI assistants · computer-vision models (YOLO, SAM). Worked on realistic production datasets — shift data, rejection counts, cycle times, sensor logs.
Outcome
Engineers replaced manual Excel-and-email reporting with automated pipelines and dashboards, learned to compute the metrics that matter directly from raw data, and left able to build predictive and vision-assisted quality workflows on their own lines — work that previously took hours, done in minutes.
Related
← All Past Trainings
The full anonymized track record of delivered corporate AI trainings.
AI for Manufacturing Program →
Full curriculum, formats, and how to book this for your team.
Corporate AI Training →
Team upskilling for engineering and L&D.
All Training →
Full catalog of AI and backend training programs.