Available for engineering roles & projects

I build systems where mistakes are expensive.

Financial data, sensitive information, and AI features people rely on — built to be correct, secure, and still working when something fails.

TW
Tobi Williams

Selected Work

Production systems built where correctness, security and reliability are non-negotiable.

01
Fintech · Tax & Financial Records
MileBack

AI-powered tax and financial management for the UK’s self-employed. I own the income, expenses and receipts area.

Rebuilt the tax calculation engine with HMRC-validated rules across multiple tax years, so annual rate changes are a one-place update instead of a bug hunt. Designed an offline-first financial data layer with idempotent sync and duplicate-safe retries, so records survive bad connectivity without loss or duplication. Built receipt scanning that reads the amount from a photo and suggests the HMRC expense category.

TypeScriptPostgreSQLLLM extractionOffline-first syncCI migrations
02
Production Platform · 12,000+ Reach
African Students’ Network

ASN started as a landing page. Its users reach it on expensive, unreliable mobile data across Africa. I lead the engineering that turned it into a platform they can rely on.

Auth, role-based access, notifications and search APIs built end to end. Offline-first caching cut data usage by 40%. Internal admin and reporting tools removed engineering from routine publishing for a 20+ person team. Now building natural-language opportunity discovery to help users find what’s relevant.

Next.jsTypeScriptFirebaseOffline-First PWACase Study →
↗
03
Multi-Tenant SaaS · Sensitive Operational Data
Data Privacy & Security Architecture

Audited and hardened 30+ FastAPI endpoints on a multi-tenant platform serving multiple commercial clients, closing cross-tenant data-leakage paths.

Enforced isolation with JWT, RBAC and PostgreSQL row-level security so no client can reach another’s data. Containerised services and set up CI/CD to GCP so releases are repeatable.

FastAPIPostgreSQL RLSDockerGCPCase Study →
↗
04
Private, On-Device AI Search
Argus

Built to prove private AI search works without the cloud.

Go handles crawling and routing; Python runs local embeddings, FAISS vector search and a local LLM via Ollama. Per-collection indexes, SQLite-backed document management and AES-256-GCM encryption at rest. Documents never leave the device.

GoPythonFAISSOllamaEncryption at restCase Study →
↗
05
Zero-to-One Builds
Startup & Product Engineering

Worked with founding teams with validated ideas and limited engineering capacity.

Shipped V1 of a due-diligence platform (subscription billing + AI document review across a 12-microservice architecture). Built KYC and investment flows for a fractional real-estate platform. Automated a daily production file pipeline that removed 20+ hours of manual work per week.

Next.jsTypeScriptPythonDockerCase Study →
↗
NOTES 3

Notes on building things people can rely on.

AI Engineering · Oct 2026
Your AI Feature Works In The Demo.
Here's Why Customers Stop Trusting It.
The problem is rarely the model. It's what's around it.
↗
Data & ReliabilityOct 2026
Why Your Customers Stop Trusting Your Numbers
Your finance product doesn't have to crash to lose a customer. It just has to show one wrong number.
Read →
StrategyJun 2026
Startups Don't Buy Code
You probably don't want a developer. You want progress.
Read →
FOR ENGINEERING TEAMS

Looking for an engineer?

I take ownership end to end: work out what matters, build it properly, make it reliable in production, and keep the team informed.

Fewer surprises, fewer half-finished features, and someone you can trust with the parts that touch money and data.

FOR FOUNDERS & BUSINESSES

Have a problem you're trying to solve?

I also take on projects. You don't need a technical spec — tell me what you're trying to make happen and what's in the way.

HOW I CAN HELP

From Problem to Working Solution

You bring what you're trying to achieve. I help you get there: working out what matters, building what's needed, and making sure it keeps working.

Sometimes that's hands-on engineering, sometimes a product built from scratch, and sometimes working out what's worth building in the first place.

01

Get your product built and shipped

From a rough idea to something your users rely on, with one person owning it from the first conversation to launch.

Product Engineering
02

Make AI actually work in your product

Not a demo that impresses once. AI features that give the right answer on the hundredth question, at a cost that still makes sense as usage grows.

AI Engineering
03

Fix what's holding you back

The release you hold your breath for, the page that times out, the data you're not sure is safe: found, fixed, and made dependable.

Backend · APIs · Data · Infrastructure · Security
04

Get time and money back

The spreadsheet someone updates by hand every Friday. The step where customers quietly drop off. Find where time, money, or customers are slipping away, and fix what's worth fixing.

Sometimes that's software. Sometimes it's automation, a better workflow, or a simpler process.

Problem Solving
THE PROCESS

How we work together

Three simple steps from problem to result. You always know what's happening and what comes next.

01

Tell me the problem

A 30-minute call. No technical spec needed, just what you're trying to achieve and what's in the way.

02

We agree the outcome

Clear scope, milestones, and price, so you know exactly what you're getting and when.

03

I build it and stay with it

Shipped, working where your users actually are, and supported after launch.

ALIGNMENT & FAQ

Everything You Need to Know About Working Together

Do you work with non-technical founders?

Yes. You don't need to arrive with a technical specification. We start with what you're trying to accomplish, what your users or customers need, and what is currently getting in the way.

I translate that into the technical decisions required and explain the important trade-offs clearly.

What kinds of problems do you work on?

There's no fixed list. I work on software products, technical systems, AI, data, infrastructure, integrations, operational problems, and other problems where my skills can create useful leverage.

If you're not sure whether I can help, tell me what you're dealing with.

Do I need to know exactly what to build?

No. You can start with the problem. We'll work out what needs to change and whether software or technology is actually the right way to address it.

What happens if the requirements change?

Requirements usually change once reality enters the picture. I work in clear iterations so we can incorporate useful information without losing control of scope, quality, or direction.

What happens if something breaks after delivery?

Production software needs to be maintained. Builds include appropriate testing, error tracking, documentation, and a defined post-launch support period. Longer-term support can also be arranged where needed.

How does pricing work?

It depends on the problem, scope, and engagement. We first understand what needs to happen. From there, I can propose the appropriate structure, such as a fixed project, milestone-based engagement, ongoing work, or another arrangement that makes sense.

How fast can we start?

Often within a week or two of our first call, depending on scope and current commitments. Small, well-defined work can usually start sooner.

What if it doesn't work out?

We agree milestones up front, and each one has to work before we move to the next. You see progress early and often, so problems surface while they're cheap to fix, and you're never locked into something that isn't working.

ABOUT

I'm Tobi Williams, a backend and AI engineer.

Most of my work sits where mistakes are expensive: money, sensitive data, and AI features people rely on.

I'm a Software Engineer at MileBack, a UK fintech, and I lead the platform at African Students’ Network. Before that I hardened a multi-tenant SaaS platform and shipped zero-to-one products with founding teams.

I enjoy problems that need more than code: understanding what matters, making good technical decisions under constraints, and building something that actually works. I'm also interested in what sits beyond engineering — businesses, markets, products.

Find something worth solving. Figure out what creates value. Build what is necessary.

Resume
CurrentlySoftware Engineer, MileBack
StackTypeScript · Python · Go · PostgreSQL
AvailableRoles · Contracts · Projects
BasedNigeria, remote (UK/EU hours)
Have something
worth solving?

Hiring an engineer, building a product, or dealing with a problem you haven't figured out yet — start with the problem.

Tell me what you're trying to make happen.