3D Neural Engine Active
Latent: 1536-D • Latency: 42ms
LANGGRAPH • QDRANT • PYTORCH
Combining university computer science studies with self taught curiosity to build practical AI tools.
I am a computer science student at Mama Ngina University College / Kenyatta University in Nairobi. Most of what I know came from deep curiosity: reading documentation, exploring open source repos late into the night, and testing things until they worked. I love building practical AI tools that do more than just write text. I focus on creating autonomous agents that can search information, answer questions accurately, and connect directly to systems like WhatsApp and M-Pesa.
I believe in simplicity and real world usefulness. A lot of AI demos look impressive in a screenshot but fail the moment a customer asks an unexpected question. I focus on writing reliable code, handling errors gracefully, and keeping the technology easy to understand for everyone. You do not need confusing buzzwords when your system simply does what it was built to do.
I focus on building practical software that solves everyday problems. Here is the primary system I built from scratch.
Autonomous Agent, WhatsApp Cloud API & M-Pesa Payments
I built Samantha as an autonomous WhatsApp assistant that makes house hunting in Nairobi simple and conversational. Users can send plain messages describing what they want, and Samantha matches them against verified listings using vector search, prompts M-Pesa STK deposits for viewings, and alerts property managers when a client is ready.
I believe tools are only as good as the problems they solve. As a computer science student and self taught developer in Nairobi, I focus on technologies that actually work in production, from smart autonomous agents to reliable payment integrations.
A basic chatbot responds to one question at a time. It predicts the next most likely words, like a helpful auto reply. But it cannot open your databases, verify information externally, or take actions on your behalf.
An autonomous agent acts like a digital team member. Given a real goal, it breaks the task into steps, searches knowledge bases, uses tools, writes code, queries APIs, triggers payments, and iterates until the job is done.
Frameworks I use to build autonomous workflows where AI agents make decisions, search data, call tools, and get real tasks finished.
Here are some of the AI tools, assistants, and software integrations I have built. I focus on practical projects that solve everyday problems rather than toy demos.
An AI assistant on WhatsApp that helps people find rental houses and apartments in Kenya. It remembers user preferences, answers questions about listings, and can trigger M-Pesa payments.
A system where specialized AI agents collaborate to finish multi-step projects. One agent gathers information, another writes code, and another checks for mistakes before giving you the final result.
A fast document search engine that lets teams ask questions about internal manuals and guides. Every answer cites the exact page reference so the AI never hallucinates.
An automated gateway that listens for Safaricom M-Pesa payments, updates customer accounts immediately, and flags duplicate or suspicious transactions.
A setup for running modern AI models completely offline on a local computer. Data never leaves your machine, making it private and free from cloud fees.
An AI assistant on WhatsApp that helps people find rental houses and apartments in Kenya. It remembers user preferences, answers questions about listings, and can trigger M-Pesa payments.
A system where specialized AI agents collaborate to finish multi-step projects. One agent gathers information, another writes code, and another checks for mistakes before giving you the final result.
A fast document search engine that lets teams ask questions about internal manuals and guides. Every answer cites the exact page reference so the AI never hallucinates.
A Telegram bot that handles customer queries and automates database checks in real time. Built with Python and running live on Telegram.
An automated gateway that listens for Safaricom M-Pesa payments, updates customer accounts immediately, and flags duplicate or suspicious transactions.
A setup for running modern AI models completely offline on a local computer. Data never leaves your machine, making it private and free from cloud fees.
An AI assistant on WhatsApp that helps people find rental houses and apartments in Kenya. It remembers user preferences, answers questions about listings, and can trigger M-Pesa payments.
A system where specialized AI agents collaborate to finish multi-step projects. One agent gathers information, another writes code, and another checks for mistakes before giving you the final result.
A fast document search engine that lets teams ask questions about internal manuals and guides. Every answer cites the exact page reference so the AI never hallucinates.
A Telegram bot that handles customer queries and automates database checks in real time. Built with Python and running live on Telegram.
An automated gateway that listens for Safaricom M-Pesa payments, updates customer accounts immediately, and flags duplicate or suspicious transactions.
A setup for running modern AI models completely offline on a local computer. Data never leaves your machine, making it private and free from cloud fees.
Interested in building an AI project, a WhatsApp assistant, or an API integration? Feel free to reach out directly or test my interactive AI assistant.
East Africa Time (EAT • UTC+3) • Open to Remote Projects
I am open to freelance client projects, contract roles, and building useful tools. Whether you have an existing system to integrate or a new project idea, let us connect.