|
sophonfinance-wq Tax Technology & AI Automation · Big 4 Tax Background · Founder, Sophon Finance Systems Tax, finance, and accounting automation — built for reviewability, evidence, and control.
sophonfinance.com
|
I'm a Tax Technology & AI Automation builder with 18+ years of accounting and tax experience, including Big 4 tax work, and the founder of Sophon Finance Systems. I build tested, reviewable automation for tax and finance workflows, combining Python systems, deterministic validation, human-gated AI review, and executive-ready reporting.
|
| Project | What it demonstrates | Stack |
|---|---|---|
| sophonfinance.com | The company site — every engine explained with animated, labeled process diagrams; all displayed figures verified against this portfolio | Static site, GitHub Pages |
| Sophon Finance Systems | Fifty runnable finance/tax automation systems, 495,064 automated tests, with evidence, validation, and reviewer-ready reporting | Python, Excel, CI, AI controls |
| Month-End Close Engine | Recurring journal entries, tie-outs, refusal controls, and close evidence | Python, pytest, JSON |
| Cash & Debt Reconciliation | GL-to-bank/lender matching with materiality classification and exception reporting | Python, pandas-style logic |
| Tax Surplus / ACB Model | Traceable surplus pools, ACB ledger behavior, and distribution waterfall logic | Python, workpapers |
| Partnership 1065 Automation | Source intake, book-to-tax bridge, 1065 mapping, K-1 preview, and review checks | Python, tax workpapers |
| Tax Workpaper Build & Carry-Down | Builds a new fiscal year's workpaper package — freezes prior-year layers, carries down allocations, constructs the schedules and the evidence tab | Python, tax workpapers |
| Triangulate AI Validation | AI separation of duties: preparer, reviewer, specialist, deterministic audit, human gate | Python, AI controls |
Every engine on sophonfinance.com has a labeled diagram like these — animated on the live site.
The operating model is intentionally control-first: source files are ingested, transformed into structured workpapers, validated through deterministic checks, and summarized for human review. AI-assisted steps are used only where they improve classification, review, explanation, or exception analysis, with final judgment kept human-gated.
Public demos use synthetic data and enterprise-safe patterns: Python, Excel-compatible outputs, Markdown/JSON evidence, tests, CI, and clear reviewer notes.
| Area | Direction |
|---|---|
| Tax and finance AI review workflows | Human-gated workflows that assist with review, validation, summarization, and approval routing |
| Tax and financial data engineering | Clean pipelines from source files to tax workpapers, reconciliations, review evidence, and management packages |
| Tax-control validation systems | Read-only checks for tie-outs, lineage, support, exception handling, and reviewer evidence |
| Reviewer and executive reporting | Short, clear summaries that help reviewers and leadership understand findings, exceptions, and next actions |
| Metric | Portfolio Signal |
|---|---|
| Repositories | Tax and finance automation systems, validation engines, tax modeling demos, and workflow documentation |
| Tests | CI-backed examples that prove behavior instead of only describing it |
| Documentation | Plain-English walkthroughs for clients, recruiters, and technical reviewers |
| Data policy | Public demos use fictional, seeded sample data only |
Visit sophonfinance.com | View the full portfolio | Start a conversation