I translate business objectives into requirements, API contracts, and data workflows for AI-native SaaS products, then trace every line back to why it matters.
↓ Download CV (PDF)I make sure engineering builds the right thing, and that stakeholders can see, trace, and verify why each piece exists. I sit between product/leadership and engineering, translating "what the business needs" into specs a dev team can build from without guessing.
| ID | Requirement | Business Objective | Test Case |
|---|---|---|---|
| FR-014 | System shall route AI queries through a fallback chain if the primary LLM provider times out. | Maintain 99% uptime for client-facing AI features | TC-014 |
| FR-021 | System shall enforce row-level tenant isolation on all client data queries. | GDPR compliance across EU client base | TC-021 |
| FR-033 | System shall log anomaly-detection flags with severity tier for reviewer triage. | Reduce manual QA review time by 30% | TC-033 |
Illustrative reconstruction for portfolio purposes: client-confidential values generalized; structure and role reflect the actual engagement.
Illustrative reconstruction for portfolio purposes: client-confidential values generalized; structure and role reflect the actual engagement.
Illustrative reconstruction for portfolio purposes, reflecting the actual delivered workflow.
| Method | Endpoint | Description |
|---|---|---|
| GET | /videos | List catalog videos with pagination and category filter |
| GET | /videos/{id}/stream | Return signed stream URL for a given title |
| POST | /users/{id}/watchlist | Add a video to a user's watchlist |
Illustrative reconstruction for portfolio purposes, reflecting the actual documented contract.
A self-directed, solo-founder BA engagement: running the full requirements-to-architecture lifecycle on an original product, the same discipline practiced at Wavess, but fully public and updated as each phase lands. Differentiated by real-time per-muscle volume tracking (MEV/MAV/MRV) and multi-method progressive overload detection, benchmarked against Hevy, Strong, Fitbod, and Tracked.
| Artifact | Key output |
|---|---|
| Charter | Business objective, scope (in/out), success criteria: artifacts must be handoff-ready and open questions must be surfaced, not silently resolved. |
| Elicitation Plan | 3 stakeholder groups mapped by power/interest; techniques matched per group (competitor-review mining, screened surveys, technical feasibility workshops). |
| Gap Analysis | Benchmarked against Hevy, Strong, Fitbod & Tracked. Confirmed real-time per-muscle volume tracking as the strongest surviving differentiator once overstated claims were corrected. |
14+ open questions logged and tracked across phases rather than resolved by assumption, each tied to the phase and decision-owner that will close it.
6 processes modeled at Level 1–2 (P0–P5) from structured BPMN specs: pools, lanes, gateways, and business rules written out before drawing. Two are shown below.
P1: Live Workout Session Logging. The core differentiator loop: set entry → RIR-weighted volume calculation → real-time feedback → overload detection (click to view full size).
P2: Authentication. Session/token validation, OAuth and email/password paths, first-time vs. returning-user routing (click to view full size).