engineering portfolio

16 projects. Built to be checked.

Governed multi-agent systems, proven where a wrong number costs the most. Every project gives the problem, the architecture, the decisions and their alternatives, and the evaluation method with its numbers.

All project data is synthetic. Individual projects document their datasets. Available walkthroughs are collected on the demos page.

CloseOps

Five governed agents with a provable no-duplicate guarantee, applied to month-end close

0duplicate postings

Read the case study

The CloseOps board: seven tasks in four columns by dependency depth, each with risk tier and gate chips. One reconciliation task is red.

LedgerGuard

An agent that can prove what it did, under seven governance checks, applied to bank reconciliation

38decisions in one reconciliation

Read the case study

FinAgent-Evals

A vendor-neutral benchmark for reliability under repetition, applied to finance reconciliation tasks

−37.8%gap between pass@1 and pass^4

Read the case study

The leaderboard: two configurations with pass@1 → pass^k columns, a consistency-gap badge on each row, and three disclosure banners above the table.

LedgerLens

An exception workbench where an uncited hypothesis is deleted in code, applied to the breaks AI matching cannot close

96%top-1 root-cause accuracy

Read the case study

LedgerLab

An open MCP server that hands any agent a realistically messy world, applied to bank and ledger reconciliation

399tests, green

Read the case study