Roman

Technical partner · idea → shipped → earning

let's fix it.

Hey Roman,
we have a
messy product
problem.

I partner with founders, agencies, and small teams to take a product from unclear idea to shipped and earning — and I own it end-to-end.

Send me the messy problem › See how I work
Available for new work
12+ Years Experience
Co-built products used by 20M+ people
Riga, Remote Europe/US
OPERATOR FILE
heyrmn
heyrmn
AVAILABLE
FOR NEW WORK
CASE FILES:
RMN · 2026 · 0042
The mess
is the job.

Why People Call Me

sound familiar?
Fuzzy product 01

The product is fuzzy, but it needs to move.

Clunky app 02

The app or site works, but users fight it.

Leaking leads 03

The site works, but people don’t convert.

Manual ops 04

Too much WhatsApp, Excel, copy-paste, and chaos.

Need a prototype 05

You need something real before committing to the build.

AI isn’t helping yet 06

Everyone talks about AI, but nothing actually gets easier.

Work & Proof

the case files

Three products, start to finish — what arrived, what I did, what shipped.

01
Case File

TonyBet Padel

Led customer-facing delivery
Client work · Intexsys
What arrived

A padel venue was being built on a €10M investment, and it needed its whole customer-facing side: a booking app, self-service kiosks, and live court screens. What existed was a high-level brief — and an opening date that would not move.

What I did

I owned everything customers and staff touch: took the React Native / Expo app from brief to soft launch, started from an initial design concept and rebuilt most of the UI from scratch as the real product took shape, designed a provisioning workflow for 12 kiosk tablets, and built a real-time tournament system that keeps 12 court screens in sync during live play. Onsite, I trained club staff on the kiosk and admin systems and turned their feedback into improvements.

What shipped

The venue opened with its systems running: 1,000+ registered users at soft launch and live tournaments across 12 courts. I stayed until people were actually using it — not until the code was merged.

MY LANE: The app, UX and final UI, 12 court tablets, admin dashboard experience, onsite rollout & staff training.
NOT MINE: Backend APIs (the backend team) · product vision (project leadership) · initial design concept (an external designer).
same-day fixes, from a laptop at the club ☺
Twelve court tablets laid out on a bench at the venue, each showing its live court screen with booking status and countdown
Court tablets · provisioning
Venue
App
1,000+
users at soft launch
12
court tablets in real-time sync
−95%
projected tablet-fleet costs
4–6h → 1h
onsite kiosk deployment
02
Case File

PFPMaker

Co-founder / Product engineer
2020 – 2025
What arrived

An idea: most profile pictures are bad, and AI could fix that in seconds. No product, no users — co-founded from zero.

What I did

Built the initial MVP and owned the product side end-to-end: frontend architecture, UX, onboarding, SEO, performance, and conversion — plus the prompt engineering behind the output quality. Led a small remote engineering team — setting priorities, reviewing code, keeping the quality bar. As a one-click profile picture grew into a full AI photo studio, I stayed its hands-on product engineer through every version.

What shipped

A launch that spread on its own — Product Hunt’s founder posted it, tech Twitter passed it around, Morning Brew put it on page one — then a free AI product used by 20M+ people: #1 Product of the Day, plus a Golden Kitty nomination.

David Sacks made his X profile picture with PFPMaker ☺
PFPMaker landing page: create AI profile pictures in seconds, 18M+ users
pfpmaker.com
David Sacks on X: “This is cool…” — quoting Ryan Hoover’s post about PFPMaker
David Sacks · Launch week
Steven Sinofsky on X: “This is great.” — quoting Ryan Hoover’s post about PFPMaker
Steven Sinofsky · a16z
20M+
people used it
55M+
profile pictures created
#1
Product of the Day on Product Hunt
5 yrs
from first MVP to mature product
03
Case File

SnapHeadshots

Co-founder / Product engineer
2023 – 2025
What arrived

A signal in PFPMaker’s traffic: people kept looking for professional headshots — and were ready to pay for them.

What I did

Validated the idea and shipped the paid MVP in three weeks. Built the mobile onboarding that gets good training photos out of any phone — camera guidance, framing masks, quality checks — feeding per-user Stable Diffusion training. Managed a dedicated AI engineer and owned the experimentation across the generation pipeline — per-user model training, preset styles, prompt engineering, and output evaluation for likeness and consistency.

What shipped

A paid product with real revenue: €54K in sales in its peak month.

Stripe dashboard: €54,524 gross volume in one month, up 375% on the month before
The peak month · Stripe
Mobile onboarding
Results
3 wks
from idea to paid MVP
€54K
sales in the peak month
2 yrs
run as a paying product
// Also Built

More Work

Other products and platforms I’ve built, beyond the case files above.

Client voice · on the outreach system
“I’ve been hiring people for 10 years and never seen this level of preparation.”
— CMO · early-stage startup
Insurance platforms
Policy, Quotes, Billing
Lead-gen sites
Funnels, Forms, CRM
SaaS dashboards
Analytics, Metrics, Admin
Landing pages
Campaigns, Acquisition
Internal tools
Workflows, Ops, Utilities
ERP integrations
Sync, Automation, APIs
Booking systems
Slots, Availability, Payments
AI search visibility
Audits, Listings, Content
Frontend & mobile // React · TypeScript · Next.js · Vue/Nuxt · React Native/Expo Backend, AI & automation // Python/FastAPI · Node.js · Supabase/Postgres · Diffusers · Replicate · LangChain · n8n · Claude Code

What We Can Do Together

you bring, I own it
Idea · Shipped · Earning
01 You bring

An Idea

We validate it, cut it to the shortest useful version, ship the MVP, and iterate until it earns.

BUILD
Proof: Case 03 · SnapHeadshots
02 You bring

A Product
That’s Stuck

I find the friction — UX, conversion, slow pages, tangled frontend — fix what’s costing you users, and get it growing again.

UNSTICK
Proof: Case 02 · PFPMaker
03 You bring

Manual Chaos

Too much copy-paste, spreadsheets, and busywork. I take it off your plate with automation and practical AI that pays for itself.

AUTOMATE
Proof: Also built · Cold Outreach
Product decisions · Delivery · Iteration after launch Owned end-to-end Not a handed-off repo

How I Work

my process
01
Understand

I dig into the real problem and what actually matters.

02
Scope

I cut noise and define the shortest path to value.

03
Plan

I turn the idea into a practical build plan.

04
Build

I use AI-assisted workflows to speed up implementation.

05
Review & Test

I review, test, refine, and make it solid.

06
Ship

You get something real — not a slide deck or half-finished codebase.

Operating Principle
AI is my multiplier, not my autopilot.

I write specs, orchestrate agents, review diffs, test behavior, own the code, and take responsibility.

Range

Frontend and product is where I’m strongest. Everything around it — mobile, backend, the visuals, the launch copy, the research — AI gets me far enough to ship it myself, without a new stack costing you weeks. When something genuinely needs a specialist, I’ll tell you — then hire that person and manage the work myself.

Before You Call

A Few Questions
How do we start? +
Send a short message, a Loom, or a voice note. I reply with whether it's a fit, a rough plan, and the next step, usually within a day. No forms, no pitch deck, no week of intro calls.
What does a typical engagement look like? +
Anything from a focused build — one flow, one automation, a first MVP — to a long-running product partnership. Either way it’s scoped tightly up front, so you always know what you’re getting and when.
How do you price? +
Per project or per week, agreed before we start — no open-ended hourly billing. You know the scope and the number going in. For the right product, I’m open to skin-in-the-game arrangements.
How does the AI-assisted workflow actually work? +
I write the specs, orchestrate agents, review every diff, and own the code. AI makes me faster across boilerplate, refactors, and research — it never ships unreviewed work. The judgment stays human.
Do you work with existing teams and codebases? +
Yes. I plug into your stack, conventions, and repo — or start clean if that's faster. Either way I leave things documented and handoff-ready, so your team isn't stuck with a black box.
What if I haven't figured it all out yet? +
That's the point. Bring it anyway — unclear scope, half-made decisions, tangled frontend. Turning that into something shippable is the whole job, not a blocker.
Impact in Numbers
20M+
Users reached
PFPMaker — AI profile-picture product
8+
Products launched
2 grew into real businesses
12+
Years experience
Frontend, product, and everything in between
Remote
Riga / Europe
Overlap with NY time zone
Ship > Talk
Current mode
Speed with clarity. No fluff.

Who I Am

The Operator

I'm Roman — a senior product & frontend engineer, 12+ years in. What I enjoy most is the early, unclear stage of a problem: working out what's actually worth building, then building it. Along the way I co-founded an AI product used by 20M+ people. Off the keyboard, I powerlift — occasionally I lift heavy things and put them back down.

Good Fit

  • › Founders, small teams, agencies
  • › Product & frontend problems
  • › Unclear scopes and half-defined problems
  • › Focused builds and longer product partnerships
  • › Practical AI & automation

Not a Fit

  • › Huge enterprise RFPs & processes
  • › Ticket-by-ticket pixel work
  • › Projects where nobody can decide
  • › Months of meetings before building
  • › Roles with no ownership and no context
Heaviest lift on record
275 kg
heaviest deadlift — lifted, then put back
no pitch deck
required
RMN-3041

Send me the messy problem

A short message, a Loom, or even a voice note — whether it's a project, a problem, or just a conversation.

hey@heyrmn.com Telegram LinkedIn