Close acceleration
Shortened post-close schedule by 5 days through standardized workflow sequencing.
Controls-first finance automation
Audit-ready finance automation.
Controls-first workflows in Alteryx, SQL, and Excel that reduce manual variance work, strengthen reliability, and accelerate close.
5 days faster post-close
30+ production workflows
2500+ annual hours saved
Controls modernization
Rule-based evaluation with audit-ready validation and traceable exception handling.
Production ROI
30+ workflows in production, reducing approximately 2500 annual manual hours.
Foundations that don't break
Repeatable ingestion and normalization patterns that produce decision-ready outputs you can actually trust.
Featured demos
Proof-of-work walkthroughs
Controlled demonstrations using synthetic data and safe examples.
Turning LLMs into a QA wingman for Alteryx
A controls-first QA workflow that narrows triage, catches breakage earlier, and produces cleaner handoff artifacts.
- Process-aware QA prompts that prioritize likely failure points first.
- Early detection for schema drift, joins, and lineage mismatches.
- Evidence-ready outputs that support audit and operating reviews.
Demonstration uses synthetic data and pseudocode only. No proprietary details are exposed.
Context Engineering for Alteryx workflow delivery
Turning dense process documentation into build-ready specifications, cleaner handoffs, and faster implementation cycles.
- Requirement translation into implementation-ready workflow blocks.
- Early schema constraints surfaced before build effort compounds.
- Documentation outputs that support operating teams after handoff.
Demonstration uses synthetic data and controlled examples only.
Retirement Planning Pipeline
A controls-first pipeline that turns a finished financial model into an evidence-backed, 80+ page retirement analysis — produced the same reliable way, every run.
- Converts a complex financial model into an 80+ page, cross-referenced analysis.
- Verification gates, resumable stages, and deterministic formatting on every run.
- Demoed with a fully synthetic persona to preserve privacy.
Demonstrated with a fully synthetic persona. No real person or account data appears anywhere.
What I build
Finance workflows built to survive review
Systems that make controls, exception paths, and operating handoffs part of the product.
Governed Data Foundations
- Ingestion to normalization to decision-ready outputs.
- Traceable transformations and consistent definitions.
Alteryx Automation Apps
- Standardized inputs, validations, and exception paths.
- Human-in-the-loop controls where required.
Finance-grade Pipelines
- Reconciliations, controls, reporting, and certification workflows.
- Deterministic, auditable, and maintainable by operating teams.
Case studies
Reusable patterns from real operating problems
Compact evidence stories for automation, QA, and handoff work.
QA wingman for Alteryx workflow review
A repeatable QA companion pattern for surfacing schema drift, lineage mismatches, and control exceptions before review cycles slow down.
Context engineering companion
A structured handoff pattern for turning process documentation into implementation-ready workflow blocks and validation notes.
Surviving LLM non-determinism in long-form generation
The architecture behind the retirement pipeline: a seven-stage gated run, a Scout Agent that gates every step, split-model routing, resumable state, and a deterministic formatting layer that make an 80-page LLM document ship reliably, every run.
How I work
Controls-first delivery model
A practical sequence for turning finance process pain into reliable automation.
- Map the process like a system.
- Surface assumptions early.
- Build deterministic workflows with explicit validations.
- Document in business language with clean handoff artifacts.
- Iterate with metrics, control testing, and feedback loops.
Lab notes
Small experiments that sharpen delivery
Notes on prompts, specs, documentation, and workflow reliability.
Prompt-guided QA
- Experiment
- Use process-aware prompts to inspect likely breakpoints before manual review.
- What happened
- The review became more focused when prompts followed the control path instead of generic code review structure.
- What I kept
- A deterministic checklist around schema drift, joins, exceptions, and handoff evidence.
Schema-first specs
- Experiment
- Start workflow design with required fields, joins, and downstream evidence outputs.
- What happened
- Ambiguity moved earlier in the process, before build effort compounded.
- What I kept
- A compact spec format that names inputs, outputs, validations, and exception owners.
Markdown style bible
- Experiment
- Use reusable Markdown patterns for SOPs, review notes, and operating handoffs.
- What happened
- Documentation stayed easier to scan and easier to version.
- What I kept
- Short sections, explicit assumptions, and action-oriented review artifacts.
Working inventory
The AI tools behind the work
A living, filterable catalog of nearly 98 AI tools, models, and architectures I've shipped, prototyped, or researched — spanning regulated enterprise finance work and independent homelab R&D.
View the inventory →About
Hello. I am Andy. And I would like to thank you for visiting my webpage.
I cut out all the baloney — what's left is the work. with apologies to Andy Kaufman
I'm a self-proclaimed geek who gets way too excited about technology (especially AI) and what it can do when it's applied first with curiosity and then discipline. I build finance automation systems that hold up under controls, audits, and real operational pressure—usually across Alteryx, SQL, and Excel.
I don't automate broken processes, but I do love untangling messy ones: mapping cause-and-effect, stress-testing edge cases, and turning “tribal knowledge” into clean, handoff-ready artifacts. If your team needs reliable automation that actually gets adopted and measurably shortens cycle time, let's talk.
Next step
Need this kind of reliability in your workflow?
Start with the process context, current control pain, and the outcome you need to trust.