[go: up one dir, main page]

Skip to content

Software and AI engineering, grounded in evidence.

You have seen the demo, and it looked good. Now someone has to put it in front of staff, customers or citizens, and answer for what happens next. Does it do the job? Where does it fail? Who is responsible when it gets something wrong?

I help organisations in Nigeria and across Africa answer those questions before they become problems. I build the system, test how it behaves and make sure someone stays accountable for it.

Portrait of Babangida Tsowa, smiling, in clear-framed glasses and a dark grey kaftan with gold embroidery.

Organisations I have worked with

Through Vorbtech Innovative Solutions and its partners. Logos are used with permission.

Three questions every AI project has to answer

Most projects meet these questions late, once people already depend on the system. It costs far less to answer them early, and they are the order my work follows.

  1. Does it do the job?

    Turning a useful idea into a working system

    A system earns its place when the people using it can finish their work without its builder beside them. I start with those people, the information they handle and the decisions they make, then build software and AI that fit how the work really happens.

  2. Where does it fail?

    Making AI behaviour easier to inspect

    A convincing answer can hide an unfinished task. You need to know what actually happened: whether the workflow finished, which checks passed, what went wrong and what a person should review. That is what my evaluation work shows you.

  3. Who stays responsible?

    Connecting governance to everyday decisions

    When someone challenges an AI system's output, they should reach a person who can do something about it. My independent research turns that idea into practical steps: named owners, proper testing, human oversight and continuing review.

What I can help with

Start wherever you are: a first decision about AI, a process that needs real software, or a team that needs to understand what it is using.

All services

Selected work

Real workflows for real institutions. Each one says what my part was, where it stands and where you can check for yourself.

All work

Who owns an AI system after the pilot?

If a public agency uses AI and it gets your application wrong, someone should be able to explain the decision and put it right. N-RAIGF and its Assurance Layer are my independent public-review papers on how Nigeria’s public institutions can make that possible, by tracing each requirement to a decision, a piece of evidence and a responsible person.

Status. Independent public-review research by Babangida Tsowa. These papers are proposals for discussion.

Read the research

How I approach a problem

You should not have to pay for a large build to find out whether an idea works. So we start small, with the work people need to do, and agree what evidence would show real progress.

  1. Understand the workflow

    We look at what people need to get done, who is involved and what gets in the way today.

  2. Define the evidence

    We agree what would show the system is working, and what would show it is not, before anything large is built.

  3. Build and test

    We build a bounded piece and test the behaviour that matters most, including how it fails.

  4. Review and hand over

    We look honestly at what remains uncertain, and leave your team able to run it without me.

That connection between practical use and evidence runs through my engineering, research and teaching.

About

I want African institutions to be able to rely on the AI they use, because they can question it. I’m a Nigerian software and AI engineer and the founder of Vorbtech Innovative Solutions, and my work so far has taken me through public-sector, healthcare and startup systems.

Alongside building products and conducting independent research, I teach practical AI and support developer education, and I’m pursuing a Masters in Information Technology, AI Specialization, at Miva University to strengthen the foundations behind it all.

Read where I am going

Latest writing

All writing

Get new writing by email

Occasional notes on building AI systems, testing them and keeping people responsible for them.

I will email you once to confirm. Unsubscribe at any time. See the privacy page.

What needs to work better?

You do not need a finished brief. Tell me about the problem you are working on and the decision you need to make, and we can work out together whether I can help.

A useful first message covers

  • The workflow or decision involved
  • Who it affects
  • What is difficult today
  • What would count as progress
Message me on WhatsApp