AI Automation & Full-Stack Engineer. in Lagos, Nigeria. I design, build and ship production AI systems end to end, from client brief to deployed product.
Most of my work is delivery: architecting the solution, integrating the moving parts, and verifying that what ships actually does what was needed. I build with AI assistance and then make sure I can explain every line of it.
alertmux: natural-hazard alerts from five official sources, normalised into one CAP-shaped schema with provenance.
Weather warnings from 59 national meteorological agencies via WMO SWIC, US alerts from NOAA/NWS, global disaster events from GDACS, earthquakes from USGS, and satellite-observed events from NASA EONET. Python, FastAPI, MIT.
It relays official warnings and never issues them. It never infers a value a source
did not supply. A missing field is null and named in unavailable_fields, and
severity codes are only translated where the mapping has been confirmed against real
CAP files. The design decisions and their reasoning are written down in
docs/DECISIONS.md,
including the ones that look like mistakes and aren't.
dictate: offline voice-to-text for macOS using faster-whisper. Local and private; audio never leaves the machine.
Backend · Python, FastAPI, Supabase (Postgres, pgvector, RLS), REST APIs AI · Claude, Codex, AI Harness, LLM app integration, RAG pipelines, MCP, voice agents (VAPI, LiveKit) Orchestration · n8n, Make.com, Trigger.dev, webhook and background jobs Frontend & deploy · Next.js, TypeScript, Astro, Vercel, Netlify, Railway, Docker
Building automation since 2024. Contract and client work, mostly private:
- Nespresso Professional AU: FastAPI keyword-research microservice on Railway, generating B2B keyword candidates with Claude, validating demand via Google Trends, and feeding an n8n pipeline. Live in the client's monthly SEO programme.
- Voice agents and RAG chatbots: inbound, outbound, reactivation and no-show flows across SMS, Instagram, Facebook and live chat.
- Multi-tenant CRM automation: Next.js and Supabase with pgvector, LangGraph agents, a human-in-the-loop approval queue, RLS, and HMAC webhook verification.
- Taught a paid six-week AI automation cohort to 10+ students, fundamentals through hands-on builds.
I also run JAMIU AI SOLUTION, where I build free tools for AI search visibility, measuring whether AI answer engines actually recommend a brand, and what to fix when they don't.
- ai.robots.txt (#267, merged): pinned a user-agent false positive as a regression test and documented why the matcher is whole-word, after checking that the bug the issue described was already fixed and saying so.
jamiusaliu090@gmail.com · jamiuaisolution.online
Open to contract and freelance work building and maintaining AI systems both globally or sponsored relocation.