// Slop rescue

Make it production-grade.

Whether your product was vibe-coded by a non-technical founder, assembled rapidly with AI coding agents, or accumulated over years by human teams, the symptoms look the same: it demos well, breaks under real users, and every change risks breaking two other things.

I rescue fragile codebases—strip out the dead weight, harden security and data integrity, add real automated guardrails, and leave software a team can actually own.

// Where slop comes from

Agent-built or human-built, the fix is engineering

// Vibe-coded & AI-assisted prototypes

Built fast with agents, now hitting production reality

You or a small team used Cursor, Claude, Lovable, Replit, or coding agents to prove a real idea and win early interest. Now real users, authentication, billing, security reviews, and silent regressions are showing up faster than prompts can patch them.

  • › Non-technical founders who built a working product and need it hardened
  • › Small teams where agent-generated code has outgrown anyone's mental model
  • › Demos that work on the happy path but fail on edge cases, concurrency or data loss
// Years of human-authored drift

Legacy codebases where every change feels risky

Slop is not exclusive to AI. Years of deadlines, agency handoffs, contractor churn, and bolted-on features produce the same outcome: duplicated logic, missing boundaries, brittle deploys, and a codebase nobody trusts enough to refactor.

  • › Core products that have slowed to a crawl under accumulated shortcuts
  • › Missing tests, leaky tenant boundaries, or fragile deployment pipelines
  • › Teams stuck choosing between an unaffordable rewrite and permanent firefighting
// What gets fixed

What a slop rescue delivers

01

Honest triage: keep, simplify or rebuild

A clear-eyed assessment of what is actually working, what is dead weight, and where the real risk lives. No gratuitous rewrites and no pretending broken foundations are fine.

02

Security, auth and tenant boundaries

Lock down authentication, authorisation, tenant data isolation, secrets handling, and input validation so the product can survive real customers and enterprise security reviews.

03

Deterministic core and data integrity

Move fragile business rules, calculations, and state transitions out of ad-hoc glue or prompts into explicit schemas, database constraints, and deterministic code you can reason about.

04

Tests, CI and anti-slop guardrails

Put real automated tests, strict types, linting, and CI gates in place so regressions fail loudly before deploy—and future human or agent edits cannot quietly erode the codebase again.

05

Production operations and observability

Reproducible environments, clean deployment pipelines, structured logging, error visibility, and sensible failure recovery so you know when something breaks before your customers do.

06

Maintainable handover

Plain architecture documentation, operational runbooks, and a codebase structured so your existing team—or the engineers you hire next—can own it without needing me on speed dial.

// How it runs

Three steps from fragile to dependable

01

Triage & risk map

Review the codebase, data model, deployment setup, and real user workflows. Separate serious production risks from cosmetic untidiness and agree a pragmatic plan.

02

Stabilise & harden

Work hands-on in the repository to fix security and data-integrity gaps, remove duplicated and dead code, add deterministic boundaries, and establish automated test coverage.

03

Guardrails & handover

Leave CI gates, clear module boundaries, and practical guardrails in place so your team—and any AI coding tools you continue using—can ship fast without slipping back into slop.

// Common questions

Frequently asked questions

Do I need an engineering team already?
No. I regularly work directly with non-technical founders as well as existing product teams. I explain trade-offs in plain English and can also help you hire or onboard the engineers who will own the product next.
Will you tell me to throw everything away and start over?
Usually no. A working prototype already encodes valuable product discovery. The default move is to salvage what works, isolate and replace the brittle parts, and only rebuild a layer when keeping it costs more than replacing it.
Does it matter whether the code was written by AI agents or humans?
Not at all. Vibe-coded prototypes and ten-year-old legacy apps fail in remarkably similar ways: weak boundaries, missing invariants, silent regressions, and low confidence in every deploy. The engineering fix is the same.
Can we keep using AI coding tools after the rescue?
Yes—in fact, AI coding tools work far better on a clean codebase. Strong types, narrow module boundaries, automated tests, and CI guardrails give coding agents clear feedback loops instead of letting them silently pile on more slop.
// Contact

Let's move your project forward

Tell me what you are building and where you want to get to. The first half-hour is free and confidential. If I am not the right person, I will say so and, where I can, point you to someone better suited.

I also consider permanent hands-on technical leadership roles, where the work is demanding and the role stays close to the code. Email is the best way to start that conversation.

Waiheke Island, Auckland · Available for remote or CBD hybrid engagements