AI & full-stack engineer in Kenya — building agents, RAG workflows, networking systems, ISP billing, cloud-backed SaaS products and automation that can survive real users. I care about practical AI: clear use cases, measured reliability, model costs, local data and systems that make it into production.
I build AI products with enough engineering discipline to leave the demo stage and work in the real world.
I’m an AI-focused full-stack engineer from Kenya building practical systems around agents, retrieval, automation and SaaS workflows. I like the part where AI meets real constraints: local data, cost, latency, trust, networking, billing operations and users who need answers they can act on.
My current direction is useful AI, not noise: agents that complete business tasks, RAG systems that cite grounded context, model choices that respect budgets and evaluation loops that show what is working. The software background still matters because AI products need backends, dashboards, queues, integrations, routers, cloud computing and deployment discipline.
“Africa should not only consume AI. We should build, test and deploy it.” — a principle I code by
I write publicly as Tim Maundu | Software & AI, exploring agents, RAG, models, cloud and how African builders can move from hype to practical deployment.
Chosen for building AI products that can be tested, shipped and improved instead of admired once in a demo.
The thread across the work: practical AI products, useful automation and software strong enough to operate.
I’m focused on AI that leaves the demo and enters the workflow: assistants that retrieve trusted context, agents that call tools safely, dashboards that expose model cost and reliability, and products that help local teams build instead of only consume AI.
Legal intelligence should be searchable, explainable and grounded. The build direction is a Kenyan-law assistant that uses retrieval, citations and careful UX so answers can be checked instead of blindly trusted.
AI-assisted employment should help people understand fit, improve applications and discover relevant opportunities. The goal is matching plus guidance, not another board of stale listings.
I’m interested in measuring what actually works: local developers trained, datasets governed responsibly, pilots moved into production and teams able to audit model failures.
Before going deeper into AI, I built and operated real SaaS, networking, ISP billing, payment, dashboard and automation systems. That background keeps my AI work tied to deployment, uptime, data flow and actual user outcomes.
Want the full list of projects & case notes?
Open projects pageShort version. The long version involves a lot of 2 a.m. deployments.
Created my GitHub workspace and began the long, rewarding climb of learning software by building it — commits before certificates.
From code into full-stack product work: PHP and MySQL backends, MikroTik networking, ISP billing, payment integration, server administration and the operational side of software — deployment, monitoring, backups, reconciliation.
Building toward AI agents, RAG products, model evaluation and cloud-backed SaaS. The goal is simple: useful AI systems that African teams can deploy, audit and trust.
Need an AI workflow, assistant, RAG system or SaaS product made dependable? I’m open to AI projects, collaborations and remote roles. GitHub and X are where I share what I build, test and learn.
~/profile open to →