Full-stack engineer for AI products

What I build

Products with a model somewhere in the loop, and the services, queues and stores that keep them honest.

• [*]CoursifiedYouTube video in, assessed course out
• [*]MBPrepExam prep on web, app and Telegram
• [*]HypatiaRaw exam papers in, verified question banks out
• [*]VectorlessRetrieval without vectors: a model reads the section tree
• [*]LorxA git forge in Rust with its own CI and reviewer
• [*]BridgeHookWebhooks to localhost, through your browser
• [*]MercalaCommerce run by chatting with agents
• [*]VoxtarVoice agent that measures the voice it hears
• [*]SelahA Bible for the terminal, on opencode's stack
• [*]CervicalLensCytology screening and genomic risk, research use
• [*]Polyglot Checkout BenchOne checkout API, ten stacks, measured
• [*]TetherLend things with a scan; one latch decides who has it

Counted, not claimed

[*]3,548 commits across 59 repositories, on 183 different days since December 2024.

Backend

30 backends, each drawn as a diagram, with the parts worth reading.

• [*]One statement decides who wins7 systems
• [*]Pipelines that resume where they stopped4 systems
• [*]Rate limits learned, like TCP2 systems
• [*]Tenancy enforced below the application4 systems
• [*]Performance as a test3 systems

Beyond code

• [*]First authorGenomic landscape of HIV-1 drug resistance in Africa, Briefings in Bioinformatics 2025
• [*]First authorCancerag, Briefings in Bioinformatics 2025
• [*]Winner, Hult Prize at UCH2025
• [*]TeachingAI in Practice: AI for Productivity: 179 registrations from 31 institutions in four countries
• [*]MedicineFinal-year medical student, University of Ibadan