About(01)
The personbehind the work
I'm a backend and AI engineer. I tend to build the parts that can't afford to be wrong: payments, e-signatures, systems that keep different companies' data apart. One person, from the database up to the AI on top.
Right now I'm building IléHQ, an AI concierge for short-let stays that lives inside WhatsApp. It's live in Lagos. A guest messages it like they would a person, and it takes them all the way from the first hello to a paid booking, on cards through Stripe or Paystack. Most of the real work is underneath, keeping one messy conversation on track until the reservation is confirmed.
Before that, about four years across fintech, proptech and edtech. I built virtual bank accounts and cross-chain transfers at Sendcoins, an e-signature engine at LuliSign, a grading system run by AI agents at Velric, and a learning platform at Digitaley Drive.
On my own time I wrote engram-dev, a memory server for AI tools that's on PyPI and scored 80% on the LoCoMo benchmark. And Revi, a self-hosted alternative to Sentry. I usually build the tools I wish I had, then become their most demanding user.
Toolkit(36)
The stack I reach for
7 disciplines, from Postgres schema to LLM agent orchestration.
Languages04
Backend07
Frontend03
AI / ML08
Data03
Payments & integrations05
Infrastructure06
Path(05)
First production role, Inframent
Built field-mapping for telecom contractors; cut search and navigation time ~25% through caching and state management.
Serverless LMS at scale, Digitaley Drive
Built an LMS on NestJS and Next.js for global cohorts, with a Paystack flow that took 70% of the manual work off people's hands.
Fintech & e-signature infra
Backend for Sendcoins' fiat-crypto platform: moving money across chains, ledgers that never double-count, KYC/AML, and a tamper-proof e-signature engine at LuliSign.
AI systems & founding IléHQ
Built a grading system on AI agents at Velric, then started IléHQ, a WhatsApp booking assistant now live in Lagos.
Open source, engram-dev
Published engram-dev, a memory server for AI tools, to PyPI. It scored 80% on the LoCoMo benchmark.
How I work
Verifiable phases, honest reporting, no big-bang reveals.
Map before I touch anything
On a new system I read it first and write down how it actually works. No blind rewrites, no guessing at the data model.
Nail the contracts
The schema, the API shapes, the money paths. Get these right and the rest of the build stops fighting you.
Ship in phases you can see
Each phase is something running, not a slide. You watch it work as it lands instead of waiting for one big reveal.
Harden the edges
Loading, error and empty states, the retries, the reconciliation. The unglamorous parts that decide whether people trust it.