Street Steel

System Concept

Street Steel was designed to bring tactical efficiency to bodyweight training scheduling. Rather than rigid preset routines, it leverages a stateless AI workout generator that maps physical traits, experience level, and available hardware (from pull-up bars to rings) directly to calisthenics training splits.

We established strict testing gates (initializing jest-expo and mocking Supabase data layers) early in the development lifecycle to ensure local client state integrity.

Prototype Interface Screens

Authentication & Splash Screen

User Stats Calibration

Equipment Selector

The Yard Dashboard

Grind Screen

Quality & Test Infrastructure

To ensure long-term codebase maintainability, we implemented an automated code quality pipeline by integrating ESLint v9 flat configs, Prettier styling configurations, and strict absolute import gates. We refactored all relative imports to absolute `@/` pathways across components, state hooks, and test files to guarantee zero visual debt.

During integration, a primary challenge was aligning external client states and mocking the Supabase data persistence layer. We configured Jest test suites utilizing `jest-expo` and resolved stub issues for the fluent database chain (such as the `upsert` queries on user profiles) to guarantee that core metrics could be verified programmatically with zero failures.

Nomenclature Matcher

To resolve the 'Nomenclature Mismatch' where the AI planner generated names different from the ExerciseDB API, we implemented a custom fuzzy search overlap matching algorithm and pre-processing name lookup. This ensures every training instruction is paired with its corresponding instructional GIF.

Ongoing Evolution

Street Steel is an ongoing engineering initiative. Our primary technical goals focus on reevaluating our training media API—as the current training GIFs lack optimal visual quality—and evolving the core authentication and payment flows into a multi-tenant subscription framework for a commercial SaaS rollout.