Projects
Each project has a write-up: the problem, how it works, the numbers, and what went wrong on the way. The code and the full notes are on GitHub.
An engine that reads a customer's message and decides what to do with it: answer, ask them to confirm, say no, or pass it to an LLM when it isn't sure. It runs on a laptop CPU, and money never moves without the customer saying yes.
0critical errors on 298 messages it had never seen
An agent for one kind of complaint, "M-Pesa took my money but my order says unpaid", from the first message to the refund. It survives crashes without paying twice, waits days for staff approval, and never shows the model a real phone number.
16/16Swahili amounts read right, after moving them from the model to code
An assistant that answers Kenyan tax questions from KRA guides and the tax Acts, with a citation for every answer. So far the work is on search: from 70% to 88% of questions finding the right section in the top five.
70% → 88%questions with the right section in the top 5
I built a fake Kenyan electronics shop and asked 11 search setups the same 71 questions in English, Swahili and Sheng, to see if Google's EmbeddingGemma 2 is worth it over what Typesense already gives you.
59% → 92%right product first, keyword search against the best setup