Case study
Cached Personnel System with Local LLM In progress
Personnel data served through a Redis cache that flushes to MySQL, employee photos in Hadoop HDFS, and a chat page backed by a language model running on the same machine.
The problem
Personnel records are read far more often than they change, and photos do not belong in a relational table. The project asks for each kind of data to live where it fits, and for the parts to still behave like one system.
What I did & decided
- Routed reads and updates through a Redis cache, with a periodic flush that writes changes to MySQL for durable storage.
- Stored employee images in Hadoop HDFS (a single-node setup) instead of in the database.
- Added a chat page backed by Llama 3.2 running locally through Ollama, so the chat needs no external API.
What came of it
- In progress — due 22 October 2026.
Next case: University Management System