$ whoami --practice

I build AI that runs
your business

An AI engineering practice for startups and operators. I find the bottlenecks in your business and replace them with AI, automation, and the lightest software that solves the problem. Shipped, not pitched.

~/status.json
building influence-v2 ▸ wk 03/08Full-stack AI platform
next_slot now ▸ opentaking discovery calls
engagement 1–8wk · fixed scope ▸ proposal < 48h
location Auckland, NZ working with clients globally
~/shipping.log live
48hr
proposal turnaround ▸ after call
1–3wk
typical build window ▸ focused scope
30d
post-launch support ▸ included
// whoami ~/about.md

One operator. End-to-end. No layers.

Rayyan Consulting is a solo AI engineering practice run by Rayyan Abzal out of Auckland. I've shipped production systems across creator-economy CRMs, payout automation, and on-chain platforms. Different domains, same engine: find the manual bottleneck, replace it with the lightest system that solves it.

Solo by design. The person who scopes your project is the person who writes the code, deploys it, and answers when something breaks. No accounts team, no junior dev getting handed the actual build. You talk to me, I write the system, you get the outcome.

~/whoami.txt
based Auckland, NZ UTC+12 · clients worldwide
focus ai_integration · automation + full-stack when it ships faster
stack typescript · python · solidity + whatever the problem needs
shipped 3 shipped · 1 building ▸ all live+ 3 self builds in the lab
capacity 2 active engagements max depth over breadth
// principles how I think about AI for businesses

AI is a tool, not a deliverable. Four principles shape every engagement.

[01] // replace, don't add

AI should replace work, not add it.

If implementing AI means more tools to learn, more dashboards to check, or more processes to follow, it's the wrong implementation. Good AI is invisible to the people whose work it replaces.

[02] // problem first

Start with the bottleneck, not the technology.

"We should use LLMs" is not a project. "This manual process eats 20 hours a week" is. Find the real bottleneck, then choose the lightest tool that solves it. Sometimes an LLM, often just well-built automation.

[03] // for the team

Build for the team, not the demo.

AI tools are only valuable if non-technical people use them every day. Ship simple interfaces a team adopts in an afternoon, not impressive demos that gather dust in a Notion doc.

[04] // hours saved

Measure in hours saved, not features shipped.

The only metric that matters is whether the business runs faster after the build. Every proposal names the manual hours we expect to eliminate, and we measure against that at handover.

// services AI · automation · software

AI-first. Full-stack when needed.

If your problem doesn't fit a chip below, the ./discovery call is free. We'll figure it out together.

Most engagements start with finding where AI or automation can replace manual work, then shipping the lightest system that does it. When the project needs a full product or platform, I build that end-to-end too.

ai_integration workflow_automation ai_native_products internal_tools full_stack_web mobile onchain

// [✓] = primary focus · examples, not a fixed menu

// selected_work 03 shipped · 01 building · 03 personal builds

Outcomes, not just code.

Each case study leads with what changed for the business. Stack and approach beneath.

personal_builds

Built for myself, to learn, scratch an itch, or test how far AI can go.

~/lab/fios self

FIOS

B2B SaaS enrichment API. Turns fragrance names into structured product data (notes, accords, longevity, AI-generated descriptions, images). Production-grade auth, rate limits, async batch enrichment.

next.jssupabaseanthropic
~/lab/aria self

ARIA

AI-powered personal stylist mobile app. Upload an outfit photo and chat with ARIA for specific style feedback based on occasion, weather, and goals. Vision API end-to-end.

react-nativeflaskanthropic
~/lab/nz-jobs self

NZ Job Finder

Scrapes Seek, LinkedIn, Indeed for grad and intern roles, scores each against your CV, generates tailored cover letters. Two-pass pipeline: seniority filter then full CV match.

typescriptplaywrightanthropic
// process discovery → proposal → build → handover

Four steps. No black boxes.

Every engagement runs through the same pipeline. You always know what stage we're at.

01 ~30min · free

discovery_call

30-minute call. Problem, goals, constraints. No pitch, just clarity on whether I can help, and if not, where to look.

02 < 48h · written

proposal + process_map

Fixed scope, fixed price, fixed timeline, plus a visual process map that lays out the plan in plain language. Non-technical founders see exactly what's being built.

03 1–8 weeks

build_phase

Weekly progress updates, live staging link from day one. Working software at every milestone. No end-of-project surprises.

04 30d included

handover

Full documentation, walkthrough session for the team, and 30 days of free post-launch support. The team owns it day one.

// testimonials from shipped engagements

In their own words.

~/testimonials/pumpdat.md client testimonial
Rayyan delivered flawless, high-quality work, faster and better than I imagined possible. Communication was smooth throughout, and he came up with smart solutions to problems before I even had to ask.

Reliable, insanely skilled, easy to work with. I'd hire him again without hesitation.

Founder
pumpdat · shipped 2025
read case study
// faq 06 common questions

Questions, answered.

I start with the problem, not the technology. If the bottleneck is a manual process that requires judgement and ingests text, images, or data. AI is usually a fit. If it's a deterministic workflow, normal automation is faster, cheaper, and more reliable. The discovery call sorts this out.
A focused automation tool or AI integration typically takes 1–3 weeks. Larger full-stack platforms can run 6–8. You get a clear timeline in the proposal before anything starts.
Yes. Most of my clients are. I handle the technical side and explain things in plain language. The proposal includes a visual process map so you can see the plan without a CS degree.
Yes. I work with existing codebases to add AI features: LLM workflows, agents, automation pipelines, without requiring a rebuild. Most engagements are additive.
Fixed-price by scope, not hourly. You know the number upfront and there are no surprise invoices. After the discovery call you'll receive a written proposal with a single price and timeline.
Happy to. Send yours, or I have a mutual NDA I can share before the discovery call if needed.
$ ./book_call --type=discovery

Let's ship something

Free 30-minute discovery call. No pitch, just a clear answer on whether I can help.