Production software for regulated industries

For founders who cannot afford
to get it wrong twice

You have either shipped with a team that could not handle the complexity, or you are about to. Centenum exists for the build that actually has to work.

Including two proprietary AI platforms in active development. We build AI infrastructure, not AI wrappers.

We do not take on projects where failure is acceptable.


Team from
Y CombinatorAgodaAmazonAnthropicToptal

Organisations our clients team members have previously worked at, not a client list.

10+
Years in fintech, government and healthcare. Not agencies. Actual production systems.
First
LGA voter verification system in Nigerian history. Government recognised.
YC W22
CTO experience at a Y Combinator company before founding Centenum.

Case studies

Shipped. In production. Still running.

Every project below is live with real consequences if it breaks. Click any to read the full story.

Fintech

PettySave MFB

A licensed bank went from zero to production

A regulated microfinance institution handling real customer money. Built the complete platform, deployed it, and handed it over running.

Read story
Government

Plateau State Government, Nigeria

The first voter verification system at LGA level in Nigeria

Electoral software that had to work on election day. It did. First of its kind in Nigerian history.

Read story
Health

DietBloom

Nutrition platform where data accuracy is not optional

Healthcare-adjacent software built with compliance as the baseline, not an afterthought.

Read story
Enterprise

Natalbond • WyseFlex • Inteluct

Enterprise platforms where downtime has a direct cost

Business automation, workforce management, enterprise operations. All live. All client-owned.

Read story

Free diagnostic

Not sure what you actually need?

Six questions. Two minutes. We will tell you exactly what kind of engagement fits your situation and what the risk is if you get it wrong.

No email required to see your result.

Clients

Companies that trusted us with the build that mattered

A licensed microfinance bank. A state electoral commission. Healthcare. Three enterprise operations.

PettySave MFB
Microfinance Banking
Plateau State / PLASIEC
Government & Electoral
DietBloom
Health & Nutrition
Natalbond
Enterprise
WyseFlex
Workforce Management
Inteluct
Enterprise Operations
Alxiom
Professional Services Network
Flobbs IN BUILD
Platform — launching soon

AI infrastructure

We build the layer most teams cannot

Most teams can call an API. Few can build the orchestration, routing, and execution architecture underneath it. This is where Centenum operates.

In Development

Kliog

Multi-tenant AI orchestration platform

Custom model routing, RAG pipelines, and a 500+ tool registry. Built for enterprises that need AI systems they own and control, not a SaaS dependency.

Model RoutingRAG PipelinesMulti-tenant500+ Tool Registry
In Development

MathExec

Formula-first modeling platform

A canvas-based math-to-model platform for ML practitioners and researchers. Write the formula, execute the model. From paper to prototype without the scaffolding.

Math-to-ModelML ResearchPaper to PrototypeFormula-first

These platforms are how we stay ahead of what clients will need next. When you engage Centenum for AI work, this is the depth you are drawing on.

What are you trying to ship?

Tell us the problem. Not the budget, not the timeline.

We respond within 24 hours. If it is not a fit, we will be direct about it.

I need a technical partner
Fintech, gov, or regulated industry
I need a Fractional CTO
Strategy, architecture, team oversight
I am not sure what I need
Take the two-minute diagnostic
← Back to all workFintech

A licensed bank went from zero to production

PettySave MFBLicensed Microfinance Bank
Live
Running in production today
Full
End-to-end delivery, zero handoff debt
MFB
Regulated financial institution

The situation

PettySave MFB had their banking licence. What they did not have was the software to run a bank. They needed an entire platform built and deployed before their regulatory window closed, handling real customer money from day one.

The margin for error was zero. A regulated microfinance institution cannot patch a bad foundation six months after launch. The system either works correctly when it goes live, or it does not go live.

A regulated financial institution cannot patch a bad foundation later. The system has to be right before the first customer touches it.

What we built

We delivered the complete Genius platform from infrastructure to handover. That meant AWS EC2 and App Runner for compute, MongoDB Atlas for the database layer, nginx for traffic handling, full SSL termination, environment management across staging and production, and a clean technical handover so their own team could own and operate the system independently.

No vendor lock-in. No dependency on Centenum to keep the lights on after delivery.

AWS EC2App RunnerMongoDB AtlasnginxSSL/TLSNode.jsCI/CD Pipeline

The result

The platform went live. A licensed microfinance bank in Nigeria runs entirely on this system today. That is the only metric that matters.

Have a build with real stakes?

Tell us what has to work.

We respond within 24 hours.

← Back to all workGovernment

The first voter verification system at LGA level in Nigerian history

Plateau State Independent Electoral CommissionCivic Infrastructure
First
LGA deployment in Nigeria
Ran
On actual election day
State
Government formally recognised

The situation

Electoral software fails in front of everyone. Voters get turned away. The result gets contested. The people responsible get named. This is not a context where you ship fast and iterate.

Plateau State needed a voter identity verification platform that could be deployed across an entire Local Government Area, handle real electoral volumes, and run reliably under field conditions.

An election is not where you discover your software does not work. There is no patch, no rollback, no recovery window. It either runs or it does not.

What we built

PLASIEC, the Plateau State Independent Electoral Commission voter verification system. Built with an offline-first architecture because field conditions in Nigerian LGAs cannot guarantee connectivity. Real-time sync when connectivity is available. Identity verification at the point of accreditation. Designed to run under pressure, on old hardware, in heat, by non-technical staff.

Identity VerificationOffline-first ArchitectureReal-time SyncField-hardened UXGovernment Infrastructure

The result

The system ran on election day. No failures reported. The Plateau State Government formally recognised PLASIEC as a landmark in Nigerian civic technology. That recognition was earned on election day, not in a boardroom.

It remains the first and only LGA-level voter verification deployment in Nigerian history.

Building something for government or civic infrastructure?

This is the context we know well.

We respond within 24 hours.

← Back to all workHealth

Nutrition platform where data accuracy is not optional

DietBloomHealth & Nutrition
Live
In production
Secure
Health data handling
Scalable
Built to grow with user base

The situation

Healthcare-adjacent software operates in a category where data inaccuracy has direct consequences for real people. DietBloom needed a platform that could handle sensitive personal health data at scale while maintaining the integrity standards that context demands.

The risk was not just technical. Incorrect nutrition data sent to users who are managing medical conditions is not a UX problem. It is a liability problem.

In health software, the consequences of getting the data wrong are not bounce rates and churn. They are user safety and regulatory exposure.

What we built

The complete DietBloom platform infrastructure: cloud architecture designed to handle sensitive health data with appropriate access controls, a mobile-ready backend built for the usage patterns of a nutrition product, and an API architecture structured for the integrations the product roadmap required.

Compliance was not retrofitted. It was the starting point.

Cloud InfrastructureData SecurityMobile-ready BackendAPI ArchitectureAccess Controls

The result

Platform live in production. Health data handled correctly. The team can build product features on a foundation that was designed for the regulatory and safety standards the category requires.

Building in health or a regulated industry?

Tell us what the compliance bar actually is for your product.

We respond within 24 hours.

← Back to all workEnterprise

Enterprise platforms where downtime has a direct cost

Natalbond • WyseFlex • Inteluct
4+
Enterprise and professional clients
Live
All in production
Owned
Clients operate independently

The situation

Enterprise operations software is not glamorous work. It is also not forgiving work. When a workforce management platform goes down, people cannot clock in. When a business automation system fails, orders stop processing. The cost is immediate and visible.

Three separate clients across different operational contexts came to Centenum with the same core requirement: software that works reliably inside their business, that their own team can own, and that does not create a permanent dependency on the vendor.

Enterprise clients do not want a vendor relationship that never ends. They want software that works and a team they can call if something goes wrong.

What we built

Natalbond: Business automation platform reducing manual operational overhead across their core workflows.

WyseFlex: Workforce management system handling scheduling, attendance, and operational reporting for a distributed team.

Inteluct: Enterprise operations platform built around their specific internal processes, not a generic template adapted for their context.

Alxiom: Complete website and digital presence for a multidisciplinary professional services network operating across North America and Nigeria. Immigration, law, investment, technology, and trade services unified under one professional platform.

In all three cases: custom-scoped, built to exact requirements, deployed to production, and handed over with documentation so the client team can operate it independently.

Business AutomationWorkforce ManagementEnterprise OperationsCustom DevelopmentClean Handover

The result

All three platforms live in production. All three clients operating their systems independently. No ongoing vendor dependency.

Building internal enterprise software?

Tell us what the system has to do and what breaks if it does not.

We respond within 24 hours.

Case studies

Every project below is in production

We do not publish case studies about things we pitched and did not ship. These are the builds where the stakes were real.

Fintech

PettySave MFB

A licensed bank went from zero to production

Complete Genius platform built and deployed for a regulated microfinance institution. AWS infrastructure, MongoDB Atlas, full handover.

Government

Plateau State / PLASIEC

First LGA voter verification in Nigerian history

Offline-first electoral system deployed at LGA level. Ran on election day. Formally recognised by Plateau State Government.

Health

DietBloom

Nutrition platform where data accuracy is not optional

Healthcare-adjacent platform built with compliance as the baseline. Sensitive health data handled correctly from day one.

Enterprise

Natalbond • WyseFlex • Inteluct

Three enterprise platforms where downtime has a direct cost

Business automation, workforce management, enterprise operations. All live. All client-owned on handover.

Have a project with real stakes?

Tell us what you are building. We will tell you if we are the right team.

We do not take on everything. That is the point.

Engagements

How we work with you

Six types of engagement. One standard across all of them: it ships, it runs, and you own it when we are done.

Production Websites
Fast, reliable websites with optional AI capabilities. Designed for conversion, reviewed for performance and security before handover.
From $3,500
Web Platforms
Portals, dashboards, booking systems, internal tools. Built for how your operations actually run, not how a template assumes they run.
From $15,000
Mobile Apps
iOS and Android, built for real usage loads. If your users depend on it daily, the reliability bar has to match that.
From $25,000
AI Infrastructure
Deep capability ↗
Custom model routing, RAG pipelines, AI orchestration, and formula-first modeling. We build the infrastructure layer, not the wrapper. Backed by two proprietary platforms in active development.
From $5,000 — see full AI capabilities
Technical Marketing
Organic traffic that compounds. SEO built for how search and AI-powered discovery actually work now, not three years ago.
Custom
Maintenance and Reliability
Security patches, performance monitoring, infrastructure updates. Your system stays current without you having to think about it.
$150 to $200 per month

Fractional CTO

Senior technical leadership without the full-time overhead. Architecture decisions, AI strategy, team structure, and investor-facing technical content. We have built and exited a YC company. That experience is what you are hiring.

Not sure what you need?

Tell us what you are trying to accomplish. We will give you an honest answer, not a proposal.

If we are not the right fit, we will say so.

Tell us the problem

Not the budget. Not the timeline. Start with what you are trying to make true that is not true yet.

We respond within 24 hours. If it is not a fit, we will be direct about it.

Where we operate
Toronto, Canada · Lagos, Nigeria
AI infrastructure & research

We build the AI layer most
teams cannot reach

Most teams can integrate a model API. Few can build the orchestration, routing, execution, and research infrastructure underneath it. That gap is where Centenum operates.

We do not sell AI features. We build AI systems.

In Development

Kliog

Multi-tenant AI orchestration platform

Built from the ground up to give enterprises real ownership of their AI stack. Custom model routing directs queries to the right model based on task type, cost, and latency. RAG pipeline architecture for retrieval-augmented generation across private knowledge bases. A 500+ tool registry that lets agents call internal and external systems without custom integration work for every new capability.

Kliog exists because enterprise AI adoption keeps hitting the same ceiling: a company integrates a model, then discovers they have no control over routing, no visibility into costs, no way to swap models without rebuilding, and no audit trail. Kliog removes that ceiling.

Model RoutingRAG PipelinesMulti-tenant Architecture500+ Tool RegistryAgent OrchestrationPrivate Knowledge Bases
In Development

MathExec

Formula-first modeling platform

A canvas-based math-to-model platform for ML practitioners and researchers. The workflow is direct: write the formula, execute the model. No scaffolding, no boilerplate, no framework overhead between the mathematical idea and a running experiment.

MathExec exists because the distance between a research paper and a working prototype is still too large. A researcher writes a loss function. Between that and a running model sits days of implementation work that has nothing to do with the research. MathExec collapses that gap.

Math-to-ModelFormula-first ModelingML ResearchPaper to PrototypeCanvas ExecutionSymbolic + Differentiable

Client engagements

AI work we take on

We are selective. We take on AI work where the infrastructure layer matters and where a cheap integration would create more risk than value.

AI Orchestration Systems
Multi-model routing, agent coordination, and pipeline architecture for organisations that need AI systems they actually own. Built on the same principles as Kliog.
Custom scoping
RAG and Knowledge Systems
Retrieval-augmented generation pipelines over private document bases, internal knowledge systems, and structured data. Built for accuracy and auditability.
From $15,000
Custom Model Integration
Integrating frontier models into production systems with proper routing, fallback logic, cost controls, and observability. Not a weekend project.
From $8,000
AI Strategy and Architecture
For leadership teams making decisions about AI investment before committing to a build. We tell you what the infrastructure requires and what it costs to get it wrong.
Fractional CTO engagement

Research

Centenum Labs

Our applied AI research arm. First arXiv publication in progress.

In Progress

StructureFind-Tabular

Empirical benchmark — compositional structure discovery in tabular data

An empirical benchmark testing how well different approaches discover compositional structure in tabular data. Subjects include PySR, gplearn, XGBoost, and frontier LLMs. Targeting arXiv publication.

Symbolic RegressionTabular DataBenchmarkarXiv

Building something that needs real AI infrastructure?

Not a chatbot. Not an API wrapper. The actual layer underneath.

Tell us what you are trying to build. We will tell you what it actually takes.