Projects

LLM League is a task-specific model evaluation system that pits AI models against each other in league and knockout tournaments. Judge models compare responses using your prompts and criteria, while the system tracks performance, consistency, position bias, and cost to help identify the best model for a specific task.

deepdive is an experimental AI-powered platform that can transform the content from a whole YouTube channel into readable, actionable knowledge through curated insights, engaging summaries, and chat assistants that let you converse with an episode or an entire channel's content.

countpesa is an android app that automatically tracks your M-Pesa transactions giving users insights into how they spend their money. It features powerful analytics on an intuitive interface that empowers users to make informed financial decisions with minimal effort. Has a 4.8 star rating on the playstore from 150+ reviews.

justTLDR is a Chrome extension that transforms YouTube videos and web articles into quick, AI summaries. It works with your API Key, or links directly to your ChatGPT, Claude, Gemini, or Grok account - letting you choose how to digest online content FOR FREE. Has 2000+ users on the chrome web store.

kalamu.app is a local-first writing tool that combines the instant snappiness of local/offline-first apps with the rich content capabilities of Notion. Store notes locally for quick load times, while syncing online for backups, cross-device access and sharing.

a privacy-focused financial analytics platform that helps M-Pesa users visualize and gain insights from transaction history. It features interactive visualizations and an AI assistant that understands natural language queries about finances—all while keeping user data private.