Portfolio demonstration showing how messy business data can be converted into a clean, standardized dataset. The workflow normalizes IDs, names, emails and categorical fields, removes duplicates, validates required data and flags records that need human review rather than silently guessing.
0
35
Portfolio demonstration of a structured data-reconciliation workflow. I created a process for comparing order and billing exports, matching records, detecting amount discrepancies, identifying invalid billing and producing a clean exception report for review. The workflow is designed around deterministic checks so results can be independently validated.
0
41
Agent Network is a Python prototype I built to test automated specialist discovery and task delegation. It matches requests with suitable providers based on capabilities, moves the task through selection and execution stages, and records the resulting transaction. I built it to explore structured multi-agent workflows and automated business processes.
0
43
AgentGuard is a Python prototype I built to explore deterministic runtime authorization for AI-agent tool calls. It evaluates requested actions and arguments before execution, blocks actions that violate policy, and records decisions in an audit log. I also built automated tests and example workflows to validate allowed and blocked behaviour.
0
25