You're taking the quarterly load prediction model from P4 and making it deliverable -- converting a notebook to a scripted pipeline, containerizing the serving endpoint in Docker, and adding prediction logging so every prediction is recorded with enough detail for the institutional-investor ESG audit trail.
The discipline skills: writing a modular scripted pipeline with YAML configuration, building a multi-stage Docker container with pinned dependencies, designing prediction logging middleware that captures timestamps, confidence scores, and raw input features, and configuring health monitoring that checks model readiness.
The AI-direction lesson: infrastructure code requires you to specify what AI won't get right on its own. AI generates FROM python:3 instead of FROM python:3.11.8-slim -- unpinned and bloated. It copies the entire project directory into the container instead of only serving dependencies. It generates monolithic 300-line scripts mixing configuration, data loading, and training. It builds logging middleware that captures request bodies without the metadata that makes logs useful. The skill here is specifying structural constraints upfront -- pinned versions, multi-stage builds, modular structure, explicit logging fields -- before AI produces code you'll have to rewrite.
Your Role
No new data, no new training. The work is infrastructure: notebook to scripted pipeline, single-machine model to Docker container, unrecorded predictions to logged audit trail.
You direct Claude Code through infrastructure work you haven't done before. Docker, scripted pipelines, and prediction logging are all new terrain. Templates and guides provide structure for each -- you fill them with design decisions.
What's New
Last time you trained a PyTorch model for Marianne Velasquez at Pinnacle Realty, encountered temporal leakage on her BMS telemetry, and learned that invisible errors in the evaluation are the most dangerous kind. The model was the deliverable.
This time the model is the starting point. The gap you're closing is between "it works on my machine" and "Carlos can run it in ten minutes." Docker containers make the environment declaration explicit — but AI generates unpinned, bloated Dockerfiles that include everything from notebooks to training code. Scripted pipelines replace notebooks with modular code and YAML configuration — but AI generates monolithic scripts with hardcoded values. Prediction logging creates an audit trail — but AI generates logging that captures request bodies without the metadata that makes logs useful. Each piece of infrastructure requires you to specify what AI won't get right on its own.
Tools
- Docker — containerization (new)
- Python — scripted pipeline, serving endpoint
- FastAPI / uvicorn — serving endpoint (familiar from P3-P4)
- YAML — pipeline configuration (new usage)
- Claude Code — AI direction
- Git / GitHub — version control
- curl — endpoint testing (familiar)
Materials
You receive:
- Marianne's follow-up email describing the operational problem
- Model artifacts from P4 (serve.py, feature pipeline, trained model, requirements, sensor schema)
- A pipeline template showing modular structure and module stubs
- A Docker guide covering container concepts, Dockerfiles, and multi-stage builds
- A logging spec template with fields to design and health check prompts
- A ticket breakdown covering setup, pipeline conversion, containerization, logging, and delivery
- A project governance file (CLAUDE.md)