Shinabaze Ajiya

Available for projects

AI automation · AI video · Websites

I build AI systems that keep working while you sleep.

Automations that answer, qualify and follow up on their own; AI-produced video with every claim checked; fast websites that sell. I run my own company on all three.

automations running in production
20+
live checks that prove they still work
25+
brands served
50+
4K explainer film, produced end to end
12:06
Shots from my 12-minute film on AI ads and the law, made in my own video pipeline.

Selected work

Systems I built and still run every day.

Four of the systems behind my own agency. Each one handles real customers, real money or real email, so each one has a guard and a test that proves the guard works.

01 Business automation

A company run by 20+ automations, with a guard on every email.

Leads, replies, payments and client briefs handled within minutes, at any hour, by one person and a set of workflows that never mail the wrong person and never break quietly.

  • Five entry points (audit form, contact form, inbound email, payment webhook, client portal) write to one customer record.
  • Every email passes one send gate in the database: opt-outs, bounces, ownership, cooldown and a daily cap. Paying clients never get marketing.
  • A central error handler and a 30-minute watchdog notice when a job stops, not just when it fails.
  • 20+ production workflows
  • 25+ live end-to-end checks
  • Each guard proven by a test that fails without it

n8n · Postgres · Supabase · AWS SES · Gemini

System diagram: A company run by 20+ automations, with a guard on every email. Architecture diagram. Steps: Five entry points (forms · inbox · payments); Classify and score (Gemini); One customer record (Supabase Postgres); Owning engine (nurture · outbound · care); Send gate (Postgres function); Email delivered (AWS SES · DKIM); Watchdog (liveness every 30 min); Central error handler (dedupe · P1 email); Refusal recorded (P1 if gate is blind); Daily digest (all green only if proven). Connections: Five entry points to Classify and score; Classify and score to One customer record; One customer record to Owning engine; Owning engine to Send gate; Send gate to Email delivered [allow]; Send gate to Refusal recorded [deny]; Owning engine to Central error handler [on failure]; Watchdog to Central error handler [reads]; Watchdog to Daily digest [daily]. GUARDS OBSERVABILITY allow deny on failure reads daily Five entry pointsforms · inbox · payments Classify and scoreGemini One customer recordSupabase Postgres Owning enginenurture · outbound · care Send gatePostgres function Email deliveredAWS SES · DKIM Watchdogliveness every 30 min Central error handlerdedupe · P1 email Refusal recordedP1 if gate is blind Daily digestall green only if proven
System diagram: A company run by 20+ automations, with a guard on every email. Architecture diagram. Steps: Five entry points (forms · inbox · payments); Classify and score (Gemini); One customer record (Postgres); Owning engine (nurture · outbound · care); Send gate (Postgres function); Email delivered (SES · DKIM); Error handler (dedupe · P1); Refusal recorded (P1 if blind); Watchdog (every 30 min). Connections: Five entry points to Classify and score; Classify and score to One customer record; One customer record to Owning engine; Owning engine to Send gate; Send gate to Email delivered [allow]; Send gate to Refusal recorded [deny]; Owning engine to Error handler [fail]; Watchdog to Error handler [reads]. allow deny fail reads Five entry pointsforms · inbox · payments Classify and scoreGemini One customer recordPostgres Owning enginenurture · outbound · care Send gatePostgres function Email deliveredSES · DKIM Error handlerdedupe · P1 Refusal recordedP1 if blind Watchdogevery 30 min
Every engine sends through one gate; failures and stalled schedules surface through the error handler and the watchdog.

02 AI agent · Support inbox

A support inbox that answers from sources, and knows when not to.

An email agent that filters, classifies, retrieves from a curated knowledge base, drafts, and lets a second model judge the draft. Only clear, low-risk answers go out on their own.

  • Suppliers, bounces and paying clients are routed to a human before any model runs.
  • Retrieval from several angles, so one phrasing never decides what the agent knows.
  • Prices in the knowledge base are generated from the same file the website sells from, and a check fails if they ever disagree.
  • 50+ node workflow
  • Judge-gated auto-send
  • Graded against a bank of real intent scenarios

n8n · Gemini · Pinecone · Gmail · Supabase

System diagram: A support inbox that answers from sources, and knows when not to. Workflow diagram. Steps: New email (Gmail poll + reconcile); Who not to answer (system · suppliers · briefs); Skipped, logged (a human owns it); Classify intent (Gemini); Multi-query retrieval (Pinecone knowledge base); Daily learning loop (corrections → KB); Draft a grounded reply (cites retrieved passages); LLM judge (rubric · revise on fail); Safe to auto-send?; Auto-send (clear, low-risk answers); Draft for a human (buying intent · doubt). Connections: New email to Who not to answer; Who not to answer to Skipped, logged [skip]; Who not to answer to Classify intent; Classify intent to Multi-query retrieval; Daily learning loop to Multi-query retrieval [updates]; Multi-query retrieval to Draft a grounded reply; Draft a grounded reply to LLM judge; LLM judge to Draft a grounded reply [revise]; LLM judge to Safe to auto-send?; Safe to auto-send? to Auto-send [yes]; Safe to auto-send? to Draft for a human [no]. skip updates revise yes no New emailGmail poll + reconcile Who not to answersystem · suppliers · briefs Skipped, loggeda human owns it Classify intentGemini Multi-query retrievalPinecone knowledge base Daily learning loopcorrections → KB Draft a grounded replycites retrieved passages LLM judgerubric · revise on fail Safe to auto-send? Auto-sendclear, low-risk answers Draft for a humanbuying intent · doubt
Filtering happens before any model runs; the judge can send a draft back for revision.

03 Web app · Payments

Sign in, pay, send a brief: the server checks everything.

A paid client portal on a static site. Checkout is tied to the signed-in account, the plan is re-read on every request, and an incomplete brief is held with the gap named instead of refused.

  • Payment links minted on the server only after the sign-in token is verified, so a payment always lands on the right account.
  • A signed, idempotent payment webhook that extends a plan instead of overwriting the days already paid for.
  • Turnaround dates from one business-day module shared by the browser and the backend, tested for parity.
  • 100+ API contract checks
  • Python ↔ JavaScript parity tests
  • Database access proven from outside

Cloudflare Pages · Firebase Auth · Supabase · Flutterwave · n8n

System diagram: Sign in, pay, send a brief: the server checks everything. Sequence diagram between Client, Static site, Pages Function, Postgres, n8n, Payment provider. 1. Client to Static site: sign in (Firebase) 2. Static site to Pages Function: checkout + ID token 3. Pages Function to Pages Function: verify token with Google 4. Pages Function to Payment provider: mint link (uid, plan) 5. Payment provider to Client: hosted checkout 6. Payment provider to n8n: signed webhook 7. n8n to Postgres: extend paid period 8. Client to Pages Function: submit brief + token 9. Pages Function to Postgres: re-read plan 10. Pages Function to Postgres: dated, or held: gap named 11. Pages Function to n8n: notify 12. n8n to Client: acknowledgement sign in (Firebase) checkout + ID token verify token with Google mint link (uid, plan) hosted checkout signed webhook extend paid period submit brief + token re-read plan dated, or held: gap named notify acknowledgement Display may trust the browser. Submission never does. Client Static site Pages Function Postgres n8n Payment provider
System diagram: Sign in, pay, send a brief: the server checks everything. Workflow diagram. Steps: Client signs in (Firebase); Verify the token (Pages Function · Google); Mint the payment link (uid + plan, server-side); Hosted checkout (payment provider); Signed webhook (n8n extends the paid period); Brief + token (plan re-read on the server); Dated, or held (the gap is named). Connections: Client signs in to Verify the token; Verify the token to Mint the payment link; Mint the payment link to Hosted checkout; Hosted checkout to Signed webhook; Signed webhook to Brief + token; Brief + token to Dated, or held. Client signs inFirebase Verify the tokenPages Function · Google Mint the payment linkuid + plan, server-side Hosted checkoutpayment provider Signed webhookn8n extends the paid period Brief + tokenplan re-read on the server Dated, or heldthe gap is named
Checkout and brief submission both start by verifying the sign-in token.

04 Growth automation · AI

Cold outreach that researches first and refuses to guess.

Sourcing, first touch and follow-ups as three stages, with enrichment, a model judge, a send gate, and reply detection that stops a sequence by reading the reply itself.

  • Prospects sourced on buying signals such as ad activity, then qualified for free before any paid lookup.
  • Drafts only with evidence: personal details counted in the body, real offers only, no invented results.
  • A model judge with failover reviews every draft; a model that does not answer sends the prospect back to the queue, never to the bin.
  • Decision-maker coverage more than doubled
  • One gate, cooldown and cap for every send
  • Judge regression bank from real drafts

n8n · Apify · Gemini · Supabase · AWS SES

System diagram: Cold outreach that researches first and refuses to guess. Workflow diagram. Steps: Buying signals (ad activity · hiring posts); Qualify for free (ICP rules · competitors); Find a decision-maker (scrape → paid lookup); Draft with evidence (3+ anchors · real offers); Model judge (failover across models); Requeue, don't reject (a timeout is not a verdict); Send gate (one owner · cooldown · cap); Cold email (touch 1); Follow-ups (touches 2–5 · 18h floor); Reply detection (attributed by content). Connections: Buying signals to Qualify for free; Qualify for free to Find a decision-maker; Find a decision-maker to Draft with evidence; Draft with evidence to Model judge; Model judge to Requeue, don't reject [no answer]; Model judge to Send gate; Send gate to Cold email; Cold email to Follow-ups [day 2+]; Reply detection to Follow-ups [stop]. 1 · SOURCE 2 · OPEN 3 · FOLLOW UP no answer day 2+ stop Buying signalsad activity · hiring posts Qualify for freeICP rules · competitors Find a decision-makerscrape → paid lookup Draft with evidence3+ anchors · real offers Model judgefailover across models Requeue, don't rejecta timeout is not a verdict Send gateone owner · cooldown · cap Cold emailtouch 1 Follow-upstouches 2–5 · 18h floor Reply detectionattributed by content
Nothing is sent without evidence and a judge’s approval.

AI video

A 12-minute film, made in my own pipeline.

A 12-minute explainer on who owns AI-made ads and who pays when they go wrong: US copyright, the platforms’ terms and the EU AI Act. Written, sourced, voiced, animated and edited in the video pipeline I built.

runtime
12:06
3840 × 2160 master
4K
frames, each placed by the transcript
21,787
loudness, broadcast-ready
−14 LUFS
  1. 01

    Claims checked before recording

    Every claim is pulled out of the script and checked against the page that should hold it. A number with no source never reaches the voice.

  2. 02

    Shots land on the words

    Narration is aligned word by word and each shot starts on the phrase it illustrates, so a re-recorded line never means re-editing by hand.

  3. 03

    Gates before anyone watches

    Loudness, colour from cut to cut and the rhythm of shot lengths are measured. A render that fails one does not ship.

  4. 04

    Cut for every screen

    One master becomes vertical Shorts that open and land on a sentence, with captions in their own band.

Four vertical Shorts, cut from the same master.

Each opens and lands on a sentence, with captions in their own band. Press play; sound is on.

  • If AI made your ad, is it still yours?

  • You’re renting your AI ad, not buying it

  • Your AI ad goes wrong. Who pays?

  • AI ads cross borders. So does the risk.

Websites

Sites I designed, built and shipped.

My agency’s site and my engineering portfolio, top of the page as a visitor sees it. Static, fast on a phone, with checkout, forms and search metadata wired in.

  • Screenshot: Agency home
    Agency homeScroll-driven brand film, built frame by frame on a canvas.
  • Screenshot: Six free audits
    Six free auditsOne form, six audits, each running an automated report.
  • Screenshot: Service landing page
    Service landing pageCopy kept as data and generated into five landing pages.
  • Screenshot: Pricing and checkout
    Pricing and checkoutSigned-in checkout, a monthly/annual toggle and a custom-plan builder.
  • Screenshot: Work and video
    Work and videoClick-to-play film with chapters and structured data for search.
  • Screenshot: Engineering portfolio
    Engineering portfolioNext.js static export, system diagrams drawn from a data model.

Services

What you can hire me for.

  • AI automation that runs unattended

    Lead capture, qualification, follow-up, reporting and hand-offs between your tools, built in n8n or code, with alerts when something stops.

    You get

    • A mapped workflow before any build
    • Error handling and a liveness check
    • A recorded walkthrough and written runbook
  • AI agents grounded in your sources

    Support and sales assistants that answer from your documents, escalate what they should not answer, and are tested against real questions.

    You get

    • Knowledge base built from your docs
    • A judge step before anything is sent
    • A test bank you can re-run
  • AI video scripted, sourced, finished

    Explainers, ads and Shorts: research, script, voice, AI visuals, edit and captions, delivered in every aspect ratio you need.

    You get

    • Fact-checked script with sources
    • Long form plus vertical cut-downs
    • Loudness and colour checked before delivery
  • Websites and funnels fast and measurable

    Landing pages, portfolios and client portals: static, quick on a phone, with checkout, forms and structured data for search.

    You get

    • Mobile-first build, tested on real device emulation
    • Search and AI-engine metadata
    • Forms wired to your automation
  1. 1

    Scope in writing

    We agree what “done” looks like, the tools involved and what I need access to, before any build starts.

  2. 2

    Build in small pieces

    You see working parts early. Each piece is tested on its own before it joins the rest.

  3. 3

    Prove the side effect

    I test what actually happened (the email delivered, the row written, the payment recorded), not a green tick on a dashboard.

  4. 4

    Hand over cleanly

    A walkthrough recording, a written runbook and a check you can run yourself to know it still works.

Clients

In their own words.

Real quotes, trimmed for length. Clients are named by role and sector, at their request.

  • We were losing up to 30% of our inbound leads because of delayed responses over weekends. DreamFoxVerse built an automated n8n pipeline that captures incoming inquiries, processes them through an AI model trained on our service documentation, and delivers structured replies in under four seconds.

    Founder · B2B services firm

  • Building a custom community tracking dashboard for our Web3 project would have taken our developers away from core protocol work for months. DreamFoxVerse built a custom portal that pulls real-time transaction data, cross-references it with Discord activity via n8n, and displays key member metrics on an easy-to-read dashboard.

    Operations Lead · Web3 project

  • We wanted a central dashboard where our clients could view project progress, upload assets, and access their deliverables without paying for a complex enterprise software subscription. DreamFoxVerse built a custom, lightweight client portal. It is clean, reliable, and has simplified our client onboarding significantly.

    Founder · Creative agency

Tools I use in production

Automation

  • n8n
  • Python
  • Node.js
  • Postgres
  • Supabase
  • Webhooks & REST APIs

AI

  • Gemini
  • Claude
  • Pinecone (RAG)
  • Tavily
  • Model judges & evals

Video

  • Remotion
  • ElevenLabs
  • Kling
  • Runway
  • HeyGen
  • ffmpeg

Web

  • Next.js
  • Cloudflare Pages
  • Vercel
  • Firebase Auth
  • Flutterwave
  • Playwright
Black-and-white portrait of Shinabaze Ajiya

About

Shinabaze Ajiya. Engineer first.

I trained as an engineer at Ahmadu Bello University and learned to build software by shipping it. For three years I have built automation, AI agents and video pipelines for clients, and I run my own agency on the same systems: its leads, support, outreach, payments and client portal are all automated.

What I bring to your project is a habit: every system gets one owner, a guard in front of anything that cannot be undone, and a check that proves it still works tomorrow.

Based in
Abuja, Nigeria
Hours
UTC+1: the whole European day, the US East morning
Languages
English
Education
B.Eng, Ahmadu Bello University

Have a system in mind? Let’s build it.

Send me a message on the platform where you found this portfolio. Tell me what you want automated, produced or built, and I will reply with questions and a plan.