A Python project series, one repo, built during the Learn2Earn AI Engineering Fellowship. Each project links to its README for the full writeup.
Across the first four projects the bugs came down to state ownership, actual types, and execution order. In Project 5, type hints with Pydantic and Pyright caught boundary mismatches before runtime — shifting effort from debugging typos to schema design.
Dictionaries, JSON persistence, separating validation from actions, ASSUMES / PROMISES / WILL NOT HANDLE
Calling an LLM API from a CLI, persisting chat history, reading an SDK instead of assuming it
Classes that own their state, serializing objects to JSON, generating PDFs
Recursive backtracking, puzzle uniqueness by counting solutions, redrawing the UI from state
Route handlers, Pydantic schema validation, relational SQLite with foreign keys & CHECK constraints, on-demand PDF generation, and live cloud deployment

A documented Python series built during the Learn2Earn AI Engineering Fellowship — real bugs, design tradeoffs, and what each build taught me.
Concept:Dictionaries, JSON persistence, separating validation from actions, ASSUMES / PROMISES / WILL NOT HANDLE
Concept:Calling an LLM API from a CLI, persisting chat history, reading an SDK instead of assuming it
Concept:Classes that own their state, serializing objects to JSON, generating PDFs
Concept:Recursive backtracking, puzzle uniqueness by counting solutions, redrawing the UI from state
Concept:Route handlers, Pydantic schema validation, relational SQLite with foreign keys & CHECK constraints, on-demand PDF generation, and live cloud deployment
Across the first four projects the bugs came down to state ownership, actual types, and execution order. In Project 5, type hints with Pydantic and Pyright caught boundary mismatches before runtime — shifting effort from debugging typos to schema design.
Now building — Project 6 (Planning)