Los Angeles, CA

I build and fix AI productswhere production mistakescost real money.

I build and fixAI productswhere productionmistakes costreal money.

Hi, I'm Aga
Aga Kadela
AI Product Engineer

B2B founders · AI SaaS · Payments · Production rescue

View portfolioGet in touch
Portfolio

A few things I've built, fixed, and rescued

01
Featured

E-Commerce Platform

Production Checkout

Checkout work for one of Europe's largest e-commerce platforms. Over a million users. Thousands of orders a day.

Under NDA. Details available on request.

1.5 years on a team where a small bad call in the wrong place could hit revenue or checkout the same day. You learn fast what a decision costs when the numbers are that loud.

ReactReduxStripeREST APIsTypeScript
02

Dotty

Launch Readiness

Invoice and payment portal built fast for validation. I reviewed it before launch, found the issues that would show up once real money started moving, and fixed them.

Fixed double-charge risk, invoice numbering under concurrency, Pacific date bugs, secrets in the repo, missing CORS on Edge Functions, and dependency debt. Shipped fixes with a severity-ranked report.

ReactSupabaseStripeAnthropic APITypeScript
Read case study →View live →
03

FunnelScout

SaaS

Multi-tenant SaaS built with AI-assisted development and hardened around the parts that matter in production: tenant boundaries, usage tracking, rate limits, and AI cost controls.

Each account has scheduled AI analysis, isolated data, and plan-based usage controls. The hard part wasn't the model. It was making sure tenant A never touched tenant B's data - and that one heavy user couldn't burn through the budget for everyone else.

Next.js 16InngestDrizzleBetterAuthStripeClaude API
Read case study →View live →
04

LA Market Report Agent

AI Agent

Monthly real estate market reporting system for a private investor in Los Angeles.

Seven data sources. Eight research tasks running in parallel. One report produced every month without someone manually stitching the research together.

HonoInngestNeonClaude APIPuppeteer
In progress

When the product gets real

Something isn't right?

The product is live, or close enough that mistakes are no longer theoretical.

Real users are in it. Payments are moving. Private data is involved. AI calls cost money now.

This is where the small things start to matter: the webhook that runs twice, the user who sees the wrong data, the flow that worked once but fails when real usage hits it.

I've seen this before.

Email me now at [email protected] and tell me what's happening. I'll get back to you today.

Email me now
About

I'm Aga Kadela. AI Product Engineer in Los Angeles.

The most expensive mistakes don't throw errors. You don't notice them until the revenue is wrong.

I learned that building checkout systems for over a million users. 1.5 years where downtime wasn't an option. You get a feel for what looks fine but will bite you later.

I bring that to everything I work on - whether I am building something new, reviewing a product before launch, or fixing something that's already live.

The question is the same either way: what actually has to work here.

I'm in Los Angeles. I work with a small number of clients at a time. Usually people who want the thinking done, not just the code written.

Full profile →

Aga Kadela

Aga Kadela

What I do

Before You Build

Get the product shape right before the code locks in the wrong assumptions: data, auth, payments, AI usage, and the first version that actually needs to ship.

Explore service →

Launch Readiness Sprint

Built it fast. Now make sure it can handle real users, payments, private data, AI costs, and the first launch that actually matters.

Explore service →

Production Rescue

Something broke and it is hitting users, money, or the workflow the business runs on. I find the fault, fix what matters first, and explain what happened in plain English.

Explore service →

Custom AI Build

You know the outcome you need. I build the system behind it - backend, data, payments, schedules, AI workflows, and the boring production parts that have to keep working.

Explore service →
Process

How I Work

Scope

I find the real problem

Most product issues show up as bugs, bad prompts, broken flows, or messy code. I start by figuring out what actually needs to be fixed, built, or left alone.

Communication

You know where things stand

You hear about what's going on before you have to ask. If something changes, I tell you - not in the next standup, now.

Judgment

I know when not to build

I'll tell you when something isn't worth building. Sometimes there's a simpler path. Sometimes we should wait.

Handover

You leave with something you can own

You leave with the code, the decisions, and the context. Fixed, explained, and not dependent on me to maintain it.

Contact

Get in touch

If you need an AI product built, reviewed, or fixed - email me at [email protected].

[email protected]LinkedIn →GitHub →
FAQ

A few things people ask

My app is already built or live. Can you help?
Yes. I work on things that are already built, already live, or stuck somewhere between prototype and production. I review the scary parts, stabilize what matters, or fix the one thing that's on fire - depends what you need.
What does an AI Product Engineer do?
I build and fix software where AI is part of the real product, not just a prototype. That means the whole stack - data, auth, money paths, integrations, and the model is only one piece. If it only works in a screen recording, it is not done.
Do you only work before launch?
No. I work before you build, before launch, and on products that are already in the wild. Stabilization, hardening, rescue - I've done all of it. Pre-launch is one part of the job, not the whole offer.
Can you work with AI-generated or messy codebases?
Yes. I'm not grading your code for style. I care whether it holds up in production. I find the risks, fix what matters on the critical path, and a full rewrite is almost never the first move.
Do I need to rebuild from scratch?
Usually not. I start with what already works and what's actually risky, then the flows that matter. A from-scratch rewrite only when it really makes sense - and I'll say so if it doesn't.
What kind of projects does Aga Kadela take on?
I work on new products and broken ones. Usually B2B teams and tech founders where the workflow is already real and mistakes are expensive.
Is Aga Kadela available for new projects?
I work with a small number of clients at a time. That's just what good work takes. If you already know there's a real problem to solve, email me at [email protected].
What industries does Aga Kadela work with?
Mostly B2B and tech-founder products. Real estate, construction, entertainment production, logistics, and AI SaaS where the workflow is already real and the software has to stay accurate.