I build reusable systems to organize information, test decisions, check evidence, and improve the work with use.
Investment Research & Decision System
The problem
Investment decisions can be distorted by incomplete research, weak assumptions, or convincing narratives.
Rigorous research across filings, earnings, valuation, risks, and alternatives takes substantial time.
I want AI to speed up the work without outsourcing the decision.
The system
Uses AI as a research and analytical assistant—not a stock picker.
Evaluates each holding in the context of the whole portfolio.
Checks important evidence and calculations before relying on them.
Tests taxes, liquidity, concentration, risk, opportunity cost, and credible alternatives.
Challenges the recommendation before I act; a companion workbook keeps changing portfolio facts separate from the durable method.
What it demonstrates: Investment research · Financial analysis · AI workflow design · Excel · Risk analysis
Current status
Built and in active use. I refine it as real research exposes better questions, stronger controls, or unnecessary steps.
See the system in practice
Example shown with fictional portfolio data to protect my financial privacy. The structure, calculations, and analytical approach reflect the system I built.
Company-Level Look-Through I built this system to combine stocks owned directly with additional exposure hidden inside funds. That lets me see how much of the portfolio really depends on each company—not just what appears on the holdings list.Whole-Portfolio Exposure The same model rolls those holdings into broader views by investment type, geography, and industry. This helps me see what the portfolio actually contains, where exposure is concentrated, and how much of that exposure comes from funds.From Analysis to Decisions I use the results to identify overlap, clarify what role each investment is playing, and surface the next questions I need to answer before making a change. The goal is not just to produce numbers—it is to build a system that supports better decisions.
AI Quality & Recommendation System
The problem
AI can sound polished while being incomplete, weakly sourced, inconsistent, or overconfident.
Important work needs more than a good prompt; it needs repeatable quality controls.
I wanted one process I could reuse across different tasks and capable AI models.
The system
Defines the objective, audience, constraints, and controlling sources before drafting.
Separates facts, assumptions, hypotheses, and recommendations.
Challenges the first approach and looks for stronger alternatives or disconfirming evidence.
Verifies important claims and tests deliverables when practical.
Runs a final multi-pass QA before output is treated as ready.
What it demonstrates: AI governance · Prompt architecture · Requirements management · Adversarial review · Quality assurance
Current status
In active use across Engineering My Best Life. I refine it when recurring failures or ambiguities expose a better control.
Cross-Model Review Workflow
The problem
One AI can miss something while sounding confident.
Asking two models is not independent review if I just choose the answer I prefer.
For high-value work, disagreement is useful evidence that something still needs to be resolved.
The system
Uses ChatGPT and Claude as separate reviewers against the same rules and evidence.
Uses each model to challenge assumptions, find gaps, test logic, and critique implementation.
Compares conclusions instead of averaging them.
Treats agreement as supporting evidence—not proof.
Uses disagreement to identify what needs more verification or a different approach.
What it demonstrates: AI evaluation · Comparative analysis · Workflow design · Validation · Critical thinking
Current status
In active use while I test whether one model is stronger overall or whether each is better for specific kinds of work.
A living portfolio
I add a system when it is ready to explain, use, and defend—and keep refining it when real use exposes a weakness.
