Senior engineering coaching for AI builders
Build your idea.
With confidence.
Build your AI software idea with a senior engineer beside you. For entrepreneurs who don’t know where to start and vibe coders who want to understand, check, and improve what they’ve built.
Hands-on guidance from David Ndungu: 25 years building software, his first AI built in 2018, and nearly two years writing code with AI.
One-to-one · Four live hours · Standard price $1,200. $600 each for the next four attendees, subject to availability.
Explore David’s projectsAI guide for your software project
Try a question
Real projects. Skills that stay with you.
Your project is
the curriculum.
Start with an idea, a business problem, or a prototype you made with AI. David helps you choose a manageable next step and understand the engineering decisions as you build. Comfort using a computer is enough to begin; programming experience is optional.
Bring a real problem
Tell David who your software should help, what you want it to do, and where you feel stuck. Start with a free 20–30 minute call to agree what’s practical.
Build one useful part
Agree on a focused outcome and a tailored agenda. Learn to plan, direct AI tools, build, and test a working part of your product in a private four-hour workshop.
Keep building
Leave with your work, reusable techniques, and next steps. Larger products may need more sessions. Shorter paid appointments and full workshops are available by agreement.
Start from where you are
Have an idea? Turn it into one useful workflow and a clear first build. Already vibe coding? Bring your prototype or a blocker. Practice tracing a problem, reviewing changes, and testing what matters with a senior engineer beside you.
Four-hour workshop: $1,200
The standard workshop price is $1,200. The next four attendees pay $600 each, with preparation included. David confirms scope, availability, and the written proposal before payment. Follow-up includes one batch of questions within seven days, up to 30 minutes of David’s time.
Workshop membership includes ongoing community access for new tools, techniques, and shared progress. Community onboarding details are confirmed with your workshop.
The code is a start.
Your judgment takes it further.
AI makes it easier to try an idea. Knowing what to ask for, what to trust, and what to check still takes practice. Sanifu gives you a place to practice those decisions on work that matters to you.
- Give your idea a finish line
- Choose one useful change. Describe who it helps and what must happen before asking an agent to build it.
- Check more than the demo
- Try the normal path, then a meaningful failure: missing input, a repeated action, or a request that should be denied. Learn what a passing check actually proves.
- Make the next change yourself
- Read the diff, explain how the feature works, and keep the decisions and next steps. Practice continuing when the first answer needs work.
What you’re paying for
Four live hours of personal guidance, preparation around your project, and feedback while you do the building. We aim for one agreed, checked piece of work, with written next steps and a method you can reuse. Deployment is included only when it fits the scope.
This is hands-on coaching for building your own software. You keep your code and accounts. Specialist ML workshops remain available by fit for experienced engineers with a concrete data or model question.
Why Sanifu?
I chose the name Sanifu, which means proper, standard, and artful in Swahili, because those are the values I embody in my engineering practice and want to share with the world.
Learn with someone
who builds.
I’m David. I built my first AI in 2018. For nearly two years, I’ve been writing code with AI, combining it with my creativity and engineering experience. Here are some of the projects I’ve created. Each has given me concrete problems to solve: context, verification, permissions, and keeping work useful after the first version.
Sanifu brings that practice to your project. You do the building, with guidance on the decisions you’ll need to make again.
Read David’s backgroundZerfoo
A Go framework for training and running ML models on your hardware. A starting point for discussing data, evaluation, and whether a model fits the problem.
Serenity
Project memory for coding agents, with a hosted preview and an open-source option. Keep useful context available when you return to the work.
Dira
A decision ledger kept in Git. Record what you chose, why alternatives were rejected, and when a decision deserves another look.
Kazi
A controller that checks a coding agent’s work against declared acceptance checks. Define what done means and inspect the evidence.
AgentGavel
An adversarial benchmark for agent governance. Test whether permission and approval controls hold when an agent is pushed to cross them.
Ajent
A place for people and agents to contribute useful knowledge and shared software. Practice checking a finding before someone else builds on it.
Explore the repositories David started in the Atlas
These are David’s projects, offered as examples of his work. Tools are chosen for your scope; using the whole collection is optional.
Is this the right next step?
Why pay when I can learn with AI and free tutorials?
Independent learning can work well. Choose Sanifu when you want someone experienced to help scope your own project, review the work with you, and explain a decision as you make it. The free call helps you decide whether that personal feedback is worth the workshop fee.
What can we accomplish in four hours?
We agree on one useful slice after the scope call: for example, one create-and-view workflow in a request tracker, or a baseline and evaluation for a suitable ML task. The aim is a checked result you understand. A complete production application, every integration, and a particular model accuracy are outside a blanket four-hour promise.
What happens after the workshop?
You keep your code and accounts, with written next steps. The pilot includes one batch of questions within seven days, up to 30 minutes of David’s time, plus ongoing community access with onboarding confirmed for your workshop. Further paid sessions are scoped separately. Any model or hosting costs are agreed before use.
Do I need programming experience or a finished idea?
No programming experience is required; you should be comfortable using a computer and willing to build hands-on. Bring an idea, a business problem, or an existing AI-built prototype. David can help you choose a small project if you are unsure where to begin. You do not need a client.
Can you help if I’m already vibe coding?
Yes. Bring the prototype and the part you do not understand or trust. Together, you can trace a bug, review a change, test a critical path, or plan the next feature. The scope call selects one useful focus; a workshop is not a comprehensive security audit or a promise that the whole app is ready for customers.
Is this a course for getting a developer job?
Sanifu’s initial focus is entrepreneurs and independent builders developing their own software. It does not provide a job-placement program, interview preparation, or promises of employment or income. An optional client project can be used for practice; you retain that relationship and any fees.
Before your workshop
Prepare for your free call.
Tell David what you want to build and where you need help. Your brief starts the conversation; it is not a purchase or a commitment.
Use the chat to explore your idea, then prepare a brief for your free call. AI can make mistakes. Chats are saved and emailed to David; please leave out secrets and private client information.
Your project brief
Review what the chat captured, make corrections, and approve the summary when it reflects your needs. David designs your agenda after the free call.
Brief approval confirms accuracy, not a purchase. David designs your agenda after a free call.
Write your own project brief
Prefer to start with your own notes? Fill in what you know, download your brief, and bring it to your free call. This form stays in your browser.