AI built your SaaS. I keep it standing while you sell.
You built a B2B SaaS with Cursor, Claude, or Lovable. It works — until it doesn’t. I step in as your fractional CTO: stabilize the architecture you already have, own the technical roadmap, and manage execution so your calendar stays free for sales. No rewrite. No equity.
Where AI-built SaaS breaks.
I read AI-generated codebases every week. The same five problems eat the founder’s calendar.
It falls over at real load
Works for 10 accounts, dies at 50. N+1 queries, zero caching, one slow endpoint locking everything.
Security was never configured
Supabase RLS wide open, service keys shipped to the browser, auth that trusts the client.
You can't see the errors
No logging, no monitoring. Every bug arrives as an angry customer email.
The schema fights every feature
Each new request ships slower than the last, because the data model was generated, not designed.
You became the support engineer
Nights in the code, days apologizing. The selling stopped.
No rewrite. Stabilize, then ship.
Audit
Fixed scope, ten days. I read your actual code — not a call, the code. You get a written report: what’s fragile, what’s exploitable, what will fail under load, and a 90-day roadmap. Yours to keep, whoever you hire next.
Stabilize
I fix the critical list myself: auth and RLS, API and webhook reliability, rate limits and abuse vectors, query performance. Your app keeps serving customers the whole time.
Roadmap & execution
I own the technical roadmap and manage execution — writing code, reviewing PRs, directing senior engineers — so you can spend 90 days talking to customers instead of the codebase.
Fee is fixed and quoted on the call, after I’ve seen the codebase.
Built for founders at one specific stage.
Worth a call if
- B2B SaaS doing $50k–$500k ARR
- 30–200 paying accounts
- Hired or about to hire a first salesperson
- Stack is Next.js, Supabase, Vercel, Stripe, or OpenAI
Not the right fit if
- No paying customers yet
- Want a full rewrite
- Crypto or Web3
- Shopping for AI strategy
Agencies rewrite. CTOs cost equity. Developers need managing. I do none of those.
An agency will quote you a three-month rewrite and hand you the same problems in a new framework. You lose six months of momentum and still own a codebase you can’t run alone.
A full-time CTO runs $180k plus equity. That’s a bet most early-stage founders can’t make without a Series A in hand.
A developer hire solves one layer down — you still have no one owning the architecture, no one setting direction, and no one to talk you out of the bad ideas at 2 a.m.
I have ten-plus years of production experience before AI existed, and I build new things with Cursor and Claude Code every week. I know where the code breaks because I read it — and I stay to fix what I find.
Production systems I’ve stabilized and shipped.
Client platforms taken from AI-built or founder-built prototypes to production, and products of my own. Every card links to a live site or a walkthrough.
Legacy Prime
Cross-platform construction-management platform for iOS, Android, and web covering estimating, CRM, field ops, and payments, with a GPT assistant that logs expenses and schedules tasks from natural language. React Native and tRPC, with Stripe billing and a Twilio AI receptionist.
OurLifeCore
Family governance and estate-planning platform with AI-drafted family constitutions, trust and estate document generation, a 2FA-protected document vault, and advisor collaboration. Multi-tenant on Next.js with 140+ API endpoints and 60+ row-level security policies.
ContractGPT
AI contract analysis with risk scoring, clause extraction, and negotiation assistance. Processes 10M+ document corpus with 1.9s response time.
AI Content Studio
An end-to-end content platform that generates social copy, marketing images, and short-form video with AI, publishes across Twitter, LinkedIn, Facebook, Instagram, and blog platforms through automated workflows, and pulls live engagement metrics from each platform's API back into a single dashboard.
AI Lead Engine
A multi-channel AI system that captures and qualifies real estate leads 24/7 across phone, SMS, and web chat. An AI assistant answers instantly, qualifies leads through natural conversation, scores them against ideal-customer profiles, and routes hot leads to the right agent while others enter automated nurture — all feeding one broker dashboard with live pipeline, agent leaderboard, and revenue analytics. Built with Next.js, Supabase, Claude, Pinecone, Vapi, Twilio, and n8n.
CHIC Concierge
"What should I wear" is a personal question, and the fashion industry answers it with infinite scrolling. A human stylist solves it well and doesn't scale past a calendar. CHIC Concierge is the stylist, scaled. Send a photo, tell it the occasion, and it builds outfit recommendations from what's already in your wardrobe and products across major retailers — computer vision reads the clothes you own, semantic search finds what pairs with them, and every recommended piece links straight out to Amazon or whichever store sells it, through affiliate links, so the look is shoppable the moment it's suggested. A recommendation product lives or dies on its matching layer, and that layer is the real work here: FastAPI on PostgreSQL, OpenAI's vision models reading garments, Pinecone powering the matching between a user's wardrobe and live retail products, Stripe for payments, Firebase underneath the app experience. It's live at digitalstylist.com.
EDVEX
AI exam-prep platform for 8+ standardized tests including SAT, GRE, MCAT, LSAT, and Bar. Adaptive practice, a voice-enabled GPT-4o tutor, AI essay grading, and spaced-repetition flashcards, built on Supabase with 35 serverless edge functions.
Sales Intelligence Platform
An end-to-end AI sales platform that takes a raw lead and autonomously researches, qualifies, calls, and books it — with zero manual work. Each lead is enriched with company data, competitor intelligence, and a tailored outreach strategy, then handed to an AI voice agent that runs a natural qualification call, applies BANT scoring, and books a meeting live on the phone. Structured call data flows back into the pipeline automatically, and a follow-up email with a scheduling link goes out the moment the call ends. Built with Next.js and TypeScript, integrating Vapi for voice, Relevance AI for enrichment, and n8n for automated follow-up — turning a process that normally eats hours of an SDR's day into a fully autonomous system.
Ten years in numbers.
What clients say.
YOU DID AN EXCELLENT, TOP-QUALITY JOB ON MY PROJECT!!! And I mean that; I have worked with many contractors in the past, and you surpassed my expectations.
Amazing programmer, highly recommended. He delivered exactly what we needed and was always available to make adjustments quickly.
Good guy. He made sure I was happy with the outcome and made changes very quickly. Would definitely work with him again.
Tools I reach for.
AI / LLM
Automation
Voice AI
Backend
Mobile
Cloud
Frontend
Testing
Ten years of mobile and web, before AI.
Production work across Cabify, Wonolo, LiveOne, Paltalk, JETSWEAT, iFoodDS, and others — iOS, Android, and web at scale before AI tools existed.
Send me what’s breaking.
Repo access or a two-minute Loom — I’ll tell you within a day what I’d fix first, and whether it’s worth fixing. Fee is fixed and quoted on the call, after I’ve seen the codebase.