ENGINEERING CASE STUDY / 05
Chronic Disease Prediction.
A serverless path from data ingestion to model inference, designed around AWS services.
PROJECT SCOPEDocumented architecture and educational source artifacts. Not clinically validated.
01 / CONTEXT
The problem worth solving.
A model needs validated inputs, repeatable data preparation, and a serving path beyond its training notebook.
Engineering objectives
- Connect ingestion to prediction.
- Prepare data for model training.
- Expose results through a simple interface.
02 / SYSTEM DESIGN
Follow the flow.
Implementation details
- Training and prediction notebooks are available.
- A Glue ETL script describes preprocessing.
- Lambda ZIP artifacts and HTML interfaces accompany sample output.
03 / TRADEOFFS
The reasoning behind the code.
Separate stages
Ingestion, transformation, training, and serving have distinct responsibilities.
Columnar handoff
Parquet connects storage and processing.
Engineering challenges
Source artifacts and an operating cloud deployment require different evidence.
Security & performance boundaries
The intended architecture uses IAM and controlled access. The portfolio collects no health information.
04 / EVIDENCE
What’s there. What’s next.
Notebooks, ETL code, Lambda artifacts, and frontend files are public. An operating AWS deployment is not verified.
Explore the source, documentation, and tests to inspect the implementation and its current boundaries.
Read the repository