Skip to content

Josemaria Aliemhe

Software Developer · AI Automation Engineer

Most of my work starts with people describing a problem in their own words, and ends with something running that their business depends on. That has meant websites, AI agents that handle customer conversations, voice agents, and the automation and infrastructure that keeps it all up. I take projects from the first conversation through to production, working closely with clients and their teams along the way.

01

Work

  1. 2026
    to present

    Seven Flags Real Estate Investments

    Nairobi real estate firm serving international property investors.

    Runs everything technical: their WhatsApp AI assistant, the databases behind it, booking and calendar integration, alerting, and the infrastructure it sits on. The assistant answers investor questions from the firm's own knowledge base, checks live calendar availability and books consultations, site visits and investor tours, records and enriches every lead, and recognises clients who return weeks later. The whole system runs self hosted on a VPS: n8n, PostgreSQL with pgvector, Docker and Traefik, with no managed database dependency.

    Result

    Runs 24/7: answers investor questions from a six document knowledge base (134 indexed chunks), books against a live calendar, and logs and enriches every lead.

    Stack
    • n8n
    • PostgreSQL + pgvector (self hosted)
    • WhatsApp Cloud API
    • Calendly API
    • OpenAI and Claude
    • Docker + Traefik VPS
    Workflow diagram. A chat trigger feeds a WhatsApp AI agent with memory, OpenAI and Claude models, a pgvector knowledge base tool, Calendly availability, and booking and lead tools writing to Postgres. Below, an ingestion chain pulls PDFs from Google Drive into the vector store.
    Agent, RAG knowledge base, booking tools and the ingestion pipeline in one workflow.
  2. 2026
    to present

    Ervin Coty

    US based author and speaker.

    Built and maintains his website and the AI content system behind it: a retrieval layer over his books and quotes so anything generated stays grounded in his actual writing, wired to a video and voice pipeline with a human approval step before publishing. Also manages the self hosted infrastructure it runs on.

    Stack
    • Next.js
    • Supabase with pgvector
    • OpenAI embeddings
    • ElevenLabs
    • self hosted n8n on Docker
  3. 2025
    to present

    Zion Health Services

    NDIS disability support provider operating across New South Wales, Victoria and Queensland.

    Built the multi page site with service pages, a multi step referral and intake flow, and search optimisation. Still manages the codebase and ships updates.

    Stack
    • Next.js
    • React
    • Tailwind
    • Vercel
  4. 2024
    to present

    Raven's Haven

    Registered NDIS provider in Victoria, Australia.

    Built the site with service pages, online referral, digital service agreements and search optimisation, and handles ongoing maintenance.

    Stack
    • Next.js
    • React
    • Tailwind
    • Vercel
02

Projects

  1. 01

    Featured

    Two sided host assistant

    whatsapp · two agents

    WhatsApp AI agent for short stay apartment hosts. One agent answers guest questions around the clock. When it cannot help, it alerts the host, takes the host's reply and relays it back to the guest in the same voice on the business number, so the guest never sees the host's personal line. Two coordinated agents with persistent memory and stateful routing between them.

    Workflow diagram. A WhatsApp trigger feeds a guest agent with memory and a chat model, a hand off branch pings the host and saves an alert, and a relay agent passes the host reply back to the guest.
    Guest agent, hand off branch and host relay in one workflow.
  2. 02

    Retail order assistant

    whatsapp · orders

    WhatsApp AI agent for Roots Beauty World, a barber and salon supplies shop in Nairobi. The agent searches the live product catalogue, quotes real prices, takes the order item by item, and only confirms once the database returns a generated order code, then tells the customer an Mpesa prompt is on the way. Order tracking works from the same chat with the order code. The pattern fits any shop that sells on WhatsApp: catalogue, orders, and status in one conversation, with nothing invented and nothing promised the system did not actually do.

    Result

    Searches a live product catalogue and creates real orders with generated order codes, end to end from chat.

    Workflow diagram. A chat trigger feeds the Asha agent with memory and an OpenAI model, plus Postgres tools for product search, order status and order creation. Disabled WhatsApp nodes show the production channel.
    Product search, order creation and order tracking against Postgres, with the WhatsApp channel alongside.
  3. 03

    Body by Alison Caroline

    web · chat and voice

    Site for a Nairobi fitness studio with class timetable, pricing and retreats, built with both a chat assistant and a voice agent so members can book by message or by phone.

  4. 04

    The Ndemi Place

    web · chat assistant

    Site for a Nairobi guesthouse with an AI chat assistant built in with Vapi, so visitors can ask about rooms, rates and event packages right on the site. A WhatsApp booking agent that checks real room availability and flags the owner when a booking needs her is built and demoed, though not currently live on the business number.

  5. 05

    Global Oil Energy and Gas

    web · catalogue and quotes

    Petroleum brokerage site with a product catalogue and quote requests.

  6. 06

    Wimly

    web · community and store

    Tennis community platform with a member area and integrated store. Events, community content, and merchandise in one place.

  7. 07

    Ehi Xpressions

    web · checkout

    Online fashion store with full Paystack checkout. Product catalogue, cart, and payment flow built end to end.

03

Skills

Web

  • Next.js
  • React
  • JavaScript
  • TypeScript
  • HTML
  • CSS
  • Tailwind
  • responsive design
  • search optimisation
  • Vercel

AI and automation

  • conversational agents
  • prompt engineering
  • retrieval augmented generation
  • vector search
  • OpenAI and Claude models
  • Claude Code
  • n8n
  • Voiceflow
  • Vapi

Backend and data

  • Node.js
  • Express
  • REST APIs
  • webhooks
  • Supabase
  • PostgreSQL
  • pgvector
  • MongoDB
  • schema design
  • WhatsApp Cloud API
  • Calendly
  • Paystack
  • M Pesa

Infrastructure

  • Docker
  • VPS deployment
  • Traefik and SSL
  • Redis
  • Git
  • domain and DNS setup
04

About

Josemaria Aliemhe

Nairobi, Kenya

Software Developer and AI Automation Engineer based in Nairobi, Kenya, open to remote.

Education

  • 2024

    Full Stack Software Development Bootcamp

    GoMyCode Lagos

  • 2021 to 2023

    University of Lagos

    Mass Communication

  • Meta

    Advanced React, Programming with JavaScript, HTML and CSS in Depth, Version Control

    via Coursera

05

Contact

josemaria14.work@gmail.com

+254 715 830829