Projects

Website and CMS FOR Baichuan Water & Power Construction LTD

Tena: AI sales assistant for Kenyan retail shops · uzatena.com Most businesses in Kenya are small, and around 60% of them are in retail. Many lose sales after the first conversation because no one follows up. Tena gives shops a simple, repeatable way to turn enquiries into repeat customers. What it does Captures every lead: walk-ins, Instagram and WhatsApp enquiries, or customers signing themselves up by scanning a counter QR code. Schedules follow-ups automatically: "I'll come back on payday" becomes a reminder on the right day. Drafts follow-ups with AI: short, local-sounding WhatsApp messages in English or Kiswahili, sent with one tap. Rewards loyalty: digital stamp cards or points tied to the customer's phone number. Includes a platform admin console: subscriptions, M-Pesa payment confirmation, suspensions, support tickets, team roles and a full log of admin actions. AI features Turns a typed note like "Faith, 0712…, wants size 42, payday Friday" into a customer record, a follow-up type and a due date. Writes each follow-up from the customer's context (chat history, purchase, why it's due), with guardrails against made-up prices or stock. Gives each shop a morning briefing on who to contact first and why. In progress: a sales-insights assistant connected through MCP (Model Context Protocol), so owners can ask questions like "What sold best this week?" Built with: React · TanStack Start · TypeScript · Tailwind · PostgreSQL (Supabase) · Better Auth · LLM APIs · Vercel

HubSpot CRM Implementation | JobTech Alliance – Mercy Corps - Migrated 4,000+ contacts from Airtable to HubSpot, including data cleaning, deduplication, and data standardization. - Configured **contact segmentation and CRM properties** to improve stakeholder organization and targeted engagement. - Built **custom dashboards and reports** to support pipeline visibility, operational tracking, and data-driven decision-making. - Integrated tools including **Zoom, Luma, and Mailchimp** to streamline event management, communications, and engagement workflows. - Configured CRM pipelines and workflows to support venture tracking, stakeholder management, and program operations. - Developed implementation documentation and delivered team training to support adoption and ongoing CRM management.

I have contributed in building and revamping HRDC website. Some of the modules that I particular built include the LMS, courses module, payment integration(Mpesa & paystack) and certificate verification.

My most complete web project. Features a comprehensive dashboard, admin panel, AI chat widget, Paystack integration, authentication & authorization, SEO, and Supabase backend.

Collaborated on designing and building an effective web presence for Cofpak, a cooperative packaging company.

I volunteered in designing and developing of a prototype for repository for HR students organisation in kenya. the site allows HR student to come together, network and identify with each other.

Meta Ads Lead Qualification and AI Voice Booking System US Cabinet Store An end to end lead handling system for a US cabinet retailer that qualifies inbound Meta ads leads by geography, calls them with a cloned AI voice, and books showroom reservations automatically. Built as four connected n8n workflows. Live in production. How it works Meta ads leads arrive through a form and land in Google Sheets, which acts as the lead database and trigger point The Nominatim API geocodes each lead address and calculates distance from the store, filtering out anyone outside the 50 mile service radius Qualified leads are routed to a Vapi outbound call using a voice cloned in ElevenLabs. Leads submitted overnight are queued and dialled the following working day Call outcomes are classified into booked, not picked, not interested, or call in future, and written back to the sheet Bookings trigger confirmation SMS through Twilio and email to both the customer and the store admin. Leads who request a specific callback time receive an SMS confirming the slot Two additional workflows let customers complete a reservation entirely over SMS or email Stack Google Sheets, Nominatim, Vapi, ElevenLabs, Twilio, Gmail

AI-powered support agent for dental clinics — streamlines patient interactions, FAQ handling, and appointment scheduling via a website widget.

Analyzing sales data with Pandas and visualizing results with Matplotlib and Seaborn.

E-commerce platform for fragrances built on Next.js and Tailwind CSS with a PostgreSQL database, deployed on Vercel, with Paystack payment gateway and both user and admin profiles.

Led a team of four in building a fully normalized bookstore database. Focused on schema design, optimization, and normalization.

Customizable AI chat widget powered by n8n and Gemini 2.5 Flash, using RAG to pull responses from a vector store. Adapts to any site theme.

Scheduling solution for VCom Networks to overcome manual shift allocation and eliminate bias. Built with React, TypeScript, and Supabase.

Exploratory data analysis and visualization of Covid-19 datasets using Python and Pandas.

WhatsApp chatbot that answers customer queries and sends automated messages, built with n8n and the WhatsApp Cloud API.

Voice agent powered by ElevenLabs, orchestrated by n8n. Handles inbound and outbound calls, supports human transfer, integrates with Vapi and RetellAI.

Property rental and home-selling platform with admin and user profiles. Explores Firebase auth with MongoDB storage.

Homes and BnB booking site built with React and Tailwind CSS backed by a RESTful Node.js API.

AI agent that learns from business support policies and FAQs to draft email replies. Supports human-in-the-loop review before sending.

Custom AI chatbot using Openrouter + Gemini 2.5 Flash, with my CV PDF as a RAG knowledge base. Embedded on this portfolio.

AI agent for laboratory management workflows, automating report generation and data collection.

Designed and built the web presence for AgoTech using Canva and Figma.

Simple receipt and invoice generation system built with Lovable (prompt-driven development) backed by Supabase.

Learning project for RESTful APIs, DB connections, and CRUD operations. Built with vanilla JS and MySQL.

Full-stack blog built with the MERN stack as a learning exercise in frontend and backend integration.