Agent Workflows
What I built
Dependency scheduling, persistent state/events, model/tool loop, schema validation, repair, concurrency and traces.
Result
Built dependency-aware scheduling, persistent workflow state, model/tool execution, schema validation and traces. The case demonstrates orchestration infrastructure and outcome grading.
Evaluation setup and selected results
A 50-task evaluation with Llama 3.2 3B captured task grades, tool-call outcomes and execution traces. The implementation demonstrates dependency scheduling, persistent state and explicit outcome grading. Task completion and task correctness are measured separately.
Compare the evaluation evidence →
About the evidence
Based on source review and saved project artifacts. Recorded measurements apply to the stated model, dataset and configuration; they are not claims of a current production deployment.