
## Summary

Deb is a debt management product focused on clarity: how debt affects cash flow, which interest rates matter, and what payoff strategies are optimal for a user’s situation.

## The problem

Debt tooling often overwhelms users with raw balances. The core experience should answer:

- “What can I safely spend?”
- “Which debt should I pay first, and why?”
- “How does this change month-to-month?”

## Constraints

- Financial data is sensitive and needs careful handling
- Strategies must be explainable (trust matters)
- Categorization and payoff logic should be testable and deterministic

## What I shipped

- Core primitives for debts, rates, and statement schedules
- Cash-flow-aware views that align with when credit card expenses actually hit
- Strategy scaffolding that supports multiple payoff methods with transparent rationale

## Architecture (high level)

- A normalized domain model for accounts + liabilities
- Calculation layer separated from UI for correctness and testability
- A recommendation layer that produces “explainable” outputs (reasons + tradeoffs)

## Outcomes

- A foundation for scaling strategy sophistication while keeping the user experience simple
- Improved ability to iterate on payoff recommendations without rewriting product flows


Debt management with cash-flow clarity, interest analysis, and payoff strategies.

Deb — smart debt management


## Summary

TrunkMatch is a marketplace for kids' hand-me-downs, matched by age, size, and family needs, with a queue that gives priority access to the best match first.

## The problem

Most marketplaces are “search first”. Here, the core value is *matching*—parents shouldn’t have to guess which listings fit their kids or who should see a trunk first.

## Constraints

- Matching needs to consider changing sizes over time (current vs outgrown).
- Sellers need a low-friction listing flow (photos + sizes + condition).
- Buyers need trust: fair prioritization, clear time windows, predictable access.

## What I shipped

- Matching logic that turns user-provided kid profiles into structured size needs
- A queue/priority model so the first-in-queue buyer gets a deadline, then it moves to the next
- Seller + buyer flows that are simple enough to use on a phone quickly

## Architecture (high level)

- Domain model for kids, sizes, trunks, and queue entries
- Deterministic ranking rules (match strength) and time-based state transitions
- Event-driven notifications (e.g., “you’re up”, “time window expiring”)

## Outcomes

- A matching-first marketplace experience that reduces search burden
- A scalable way to enforce fairness and prioritization without manual moderation


A marketplace for kids' hand-me-downs, matched by age, size, and family needs.

TrunkMatch — marketplace matching + queueing for hand-me-downs


## Summary

With Light is a DBT-first emotional wellness product that helps users track emotions and situations, then guides them to the DBT skills and exercises they need in the moment.

## The problem

When someone is dysregulated, they need help quickly. The product experience needs to be:

- fast (few taps to log)
- guided (clear next steps)
- structured (skills are evidence-based, not generic advice)
- safe (sensitive data, careful UX)

## Constraints

- The system must support a growing library of DBT skills/exercises without hard-coding flows.
- “Personalization” needs to be explainable and controllable (avoid random-feeling recommendations).
- Privacy expectations are high (users are logging sensitive situations).

## What I shipped

- A skills library that can power multiple surfaces (browse, search, “recommended for you”, guided exercise flows)
- A matching layer that maps user state (emotion + context + distress) → relevant skills and next steps
- Guided exercises that break down DBT tools into step-by-step interactions

## Architecture (high level)

- A content model representing skills, exercises, and prompts as composable modules
- A recommendation/matching subsystem with deterministic fallbacks and guardrails
- Telemetry designed around product iteration (drop-offs, completion rates, time-to-first-tool)

## Outcomes

- A product foundation that supports adding new DBT tools quickly and consistently
- Clear separation between content, matching logic, and UI surfaces (improves maintainability and experimentation)

## What I’d do next

- Add lightweight experimentation (A/B) around tool matching and exercise completion
- Expand “insights” to help users see patterns while keeping the UX simple and non-judgmental


A DBT-first emotional wellness product: fast tracking, personalized skill recommendations, and guided exercises.

Projects

Deb — smart debt management

TrunkMatch — marketplace matching + queueing for hand-me-downs

With Light — DBT tools + guided AI coaching