I build AI systems that know when to act — and when to ask first.
Automation and AI engineering for real operations: agent orchestration, RAG, and human-in-the-loop approval gates on top of n8n, Python, and production data stores. Not dashboards that display information — systems that act on it, safely.
About
I'm an AI and business process automation engineer based in Lagos, working across the full stack of an agentic system: intake, retrieval, reasoning, tool-calling, and — critically — the guardrails around what an agent is allowed to do without a human signing off.
My background is mechanical engineering (HND, Yaba College of Technology), which is where the instinct for deterministic logic and failure modes comes from. I moved into automation and AI engineering through Microtech Africa's automation and agentic AI programs, and now build production-style systems for Nigerian SMEs while working remotely as a research assistant for BrandPinpoint (US).
The thread across my projects: agents that don't just answer questions, but take real actions — book a viewing, raise a purchase order, approve a refund — with a validation and escalation layer sitting between the model's decision and anything consequential actually happening.
ORCHESTRATION
AI / RETRIEVAL
DATA / OPS
Systems
Six systems spanning real estate, e-commerce, retail inventory, healthcare scheduling, and internal tooling. Teal marks a fully autonomous step; amber marks a step gated behind human approval.
A property inquiry and viewing-scheduling assistant for Pride Rock Properties. A Supervisor Agent routes each Telegram message to a RAG inquiry tool or a scheduling tool, with slot confirmations and edge cases gated behind Slack approval before anything is booked.
An automated reordering system for a multi-location Lagos supermarket chain. Deterministic stock math flags reorder candidates, an AI reasoning layer recommends a supplier, and a two-stage human approval chain (Lead Merchandiser, then Branch Manager) signs off before any purchase is created.
A full order management and fulfilment pipeline for a fictional Nigerian fashion brand, built as capstone lead across a six-workflow architecture: WhatsApp order parsing, payment capture, carrier dispatch, and AI-driven CRM scoring and reporting.
A medical appointment booking system with AI-based doctor-specialty triage, slot-conflict resolution, and a multi-pipeline architecture covering confirmations, reminders, daily doctor digests, and post-appointment follow-up.
A production-pattern customer support agent for an electronics retailer, with persistent cross-session memory, RAG over a product knowledge base, and a secondary extraction step that writes structured customer facts to Postgres.
A research assistant agent built for automation engineers and consultants — surfaces company, industry, and job-posting signals to speed up client and lead research.
Working stack
ORCHESTRATION & LOGIC
AI & RETRIEVAL
DATA & MESSAGING
Let's talk
Open to AI Engineer and automation consultant roles, and to freelance builds for teams that want agents doing real work — not just answering questions.