Back to portfolio

Case Study

AI Log Analyzer

LLM Infrastructure / API project

FastAPI service that validates raw logs and produces structured diagnostic output with OpenAI to support incident review and triage.

RepositoryMore projects

Stack

PythonFastAPIOpenAIPydanticDocker

Problem

Incident review often loses time in unstructured logs, repeated manual triage and inconsistent diagnosis output.

Approach

The project exposes a FastAPI backend that sends logs through validation layers and OpenAI-powered structured diagnosis flows.

Key Decisions

  • Use Pydantic models to keep inputs and outputs explicit.
  • Treat the LLM as one component behind API and validation boundaries.
  • Design for retries, timeouts and predictable API behavior.

Tradeoffs

  • LLM output needs guardrails, so structured responses are favored over free-form answers.
  • The portfolio page avoids claiming operational incident metrics that are not published in the repository.