Skip to content
Kenny Gazko
20+ years in tech & support · 5+ years in operations

I build the platform, automate it, and get out of the way.

I replace a company's SaaS stack with systems it owns, then automate them until nobody has to think about them.

Would I be able to help you grow?Try 6 live demosSee how it worksHire me on Upwork

Monthly tooling spend, one client

$15,000$500

I replaced the stack with software the company owned outright.

“He's not just completing work, he's improving how the work gets done.”

Porfirio, Chief Operating Officer, FunnelPros.ai / Nonstop Automation

3 → 13
headcount scaled while building the systems under it
90% → 10%
dispute losses, inside two months
12
systems designed, built and shipped
OnboardingCustomer supportSales pipelinesHR recordsAccountingInventorySchedulingTicketingReportingForecastingDataset pipelinesInternal commsDocument workflowCompliance logsKPI dashboardsAlertingApprovals & sign-offAsset managementTime trackingQuality assurance
What I actually do

Complete product and operations automation

Most companies run on nine subscriptions that don't talk to each other, and somebody in the middle manually copying between them. I replace that with one system the business owns, built around how the work actually gets done, not how a vendor assumed it would.

Frontline work feeds an owned platform, which removes busywork and gives leadership a live viewFRONTLINESupportSalesFloor & opsONE PLATFORM · YOURS, NOT RENTEDOnboardingTicketingDatasetsApprovalsTime trackingReportingSchedulingInventoryAlertingLEADERSHIPLive positionProjectionsExceptionsworkcaptured oncebusywork removedliveno waiting for reportsNo re-entry between tools · no per-seat licences · no vendor holding your data

I start at the bottom of the chain

Not with the org chart, with the person doing the job. They know exactly where the process leaks, and they're almost never asked. Their pain points become the spec.

The system runs itself

Automation isn't a dashboard someone has to remember to open. Work gets captured where it already happens, routed without a human deciding, and escalated only when something genuinely needs a person.

You own it outright

In-house means no per-seat bill that grows with headcount, no renewal negotiation, and no vendor holding your operational data hostage. The 200th user costs the same as the 20th: nothing.

Every department, one system

Built as modules, so you can start with the one that hurts most.

Customer support

Ticketing, routing, SLAs, macros, escalation paths, satisfaction tracking.

Sales

Pipeline, quoting, follow-up automation, forecasting that reflects reality.

HR & onboarding

Hiring flow, document collection, training tracks, time and scheduling.

Accounting

Invoicing, reconciliation, expense capture, payroll inputs, audit trails.

Inventory & assets

Stock, equipment, vendors, purchase approvals, reorder triggers.

Reporting & projections

Live KPIs, forecasting, exception alerts, surfaced rather than requested.

This isn't theoretical

As COO at Next Wave Services I built an operation that ran without daily oversight, and then moved from full-time to part-time to advisory as the system matured. The automation worked well enough that the company stopped needing me in it every day. That is the entire pitch, already proven once.

I spent 20+ years in technology and customer support and 5+ years in operations before I ever pitched a system to anyone. I know what a queue feels like at 4pm on a Friday, and which “small” process change quietly costs a team six hours a week. That's the difference between software built for a business and software built by someone who has actually run one.

See the full track record
Runs wherever you need it

iOS, Android, macOS, Windows, Linux, the browser, the tablet bolted to the wall in the warehouse. A system that only works at a desk is only half a system, so the delivery target is a decision we make together rather than a limitation you inherit.

Smart fridge, toaster, whatever it is. If it has a screen and a network stack, I will build for it.

Engagements

I work with all budgets, with stipulations: scope moves to fit what's actually available, and I'll tell you up front what a given budget can and can't buy.

No problem is too small or too big. Bring me the idea; I'll turn it into a running system the employees and the employer both trust.

How it works

I take the support role first, then build the system

Most people are hired to fix operations and start with a stakeholder interview at the top. I start at the bottom, on the phones, doing the job, because the failures that matter are invisible from a conference room. Then, and only then, I build.

A six-step ascent from working the frontline job up to the CEO, then buildingWork the jobWatch the painFind the weak linkTalk to the teamReach the CEOBuildI start heremost consultants start here

The spec writes itself if you're standing in the right place

By the time I propose anything, I've personally hit the same wall your team hits, watched a customer get frustrated by it, and heard from the person who's been quietly working around it for a year. That's not research. That's just having done the job.

Happy employees, happy customers, happy owner

Those three are the same problem viewed from three heights. When the frontline isn't fighting the tooling, the customer feels it immediately, and the owner sees it in the numbers. I bridge those gaps so the whole thing runs at 100%.

01

I work the actual job

Not a discovery workshop, but a normal Customer Support role, on the rota, taking real tickets. You cannot see how a system fails from a meeting about the system.

02

I watch the pain happen live

Customer frustration in the moment it occurs, not summarised in a quarterly report three months later with the context stripped out.

03

I find the weak link and document everything

Every workaround, every re-entry, every place the process quietly depends on one person remembering something. Written down as it's found.

04

I talk to the employees

Their personal pain points, not the official ones. The thing they've stopped complaining about because they assume it can't be changed is usually the biggest win available.

05

I work my way up to the CEO

And ask the question that matters: what can you see, and what can't you? The gap between what leadership thinks is happening and what's actually happening is where the system gets designed.

06

Then I build

By this point the spec isn't a guess. It's a list of real failures with names attached, and I already know who has to live with the result.

Then I build

Five habits, with the receipt for each.

01

Decide what's expensive to reverse

I separate the choices that are cheap to change from the ones that aren't: data model, auth boundary, deployment target. Those get decided deliberately. Everything else can be discovered while building.

Claw Gunners shipped a browser prototype first, purely to settle design questions before committing to an engine.

02

Make it usable at the end of every phase

There is always something running. No three-week stretches with a broken build and a promise. Each phase ends with a thing that can be shown, used and judged by the people who will live in it.

SuperTurkBot ran 37 numbered phases, each with its own changelog entry and shippable state.

03

Automate until the system doesn't need me

The goal isn't a tool that requires a specialist to operate. It captures work where it already happens and escalates only when a human is genuinely required, including when that human was me.

At Next Wave Services the operation ran without daily oversight, which is why the COO role moved to part-time and then advisory.

04

Write the decisions down

Design guidelines, schema documents, migration histories, QA reports with root causes. Not ceremony. It is what lets someone else pick the work up, and what stops me relitigating a decision I already made.

SuperTurkBot runs a hand-rolled migration runner with drift history, so every schema change is reversible and auditable.

05

Cut scope when the evidence says to

The hardest call is removing something you already built. If a dependency costs more than it returns, it goes. Sunk cost is not an argument.

A.V.Y.N. ran on Matrix until the federation complexity stopped paying for itself, then it was removed entirely.

Where I fit best

The earlier you are, the more leverage there is, but nobody gets turned away for being the wrong size.

Startups

Best fit

Nothing is entrenched yet. I can set the operating system up correctly the first time instead of unpicking three years of workarounds.

Small businesses expanding

Best fit

You're about to double and the manual process that worked at five people is about to break at fifteen. This is the moment it's cheapest to fix.

Medium-sized businesses

Also welcome

You're not left in the dark. I come in, assess the situation, give you my honest read, and we take it from there, whether that's a full build or one department.

In-house first

I'm not a fan of leaning on outside APIs. If I can build it in-house, I do. It costs less, it doesn't break when someone else changes their pricing, and you own it. When an external service genuinely is the right call, I integrate it in the way that keeps the running cost low and the operation clean, rather than wiring your business to a meter.

No idea is too small

Every project on this site started as a small annoyance I wanted to stop having. Each one grew into something that could move an industry standard. The “minor” irritation your team has stopped mentioning is usually the thread worth pulling. Bring me that one.

Case study · TutorMax · Oct 2025 – May 2026

$15,000 a month in tooling, replaced by $500

A startup was running its whole operation on rented software. Over that engagement I replaced the stack with systems the company owned, then automated the part everyone hates: writing things down. What follows is what I built and what it did while I was there.

Before

$15,000

per month, in subscriptions

After

$500

infrastructure upkeep + GoHighLevel

Removed from opex

~$174k

per year, recurring

What it replaced

  • ClickUp doing double duty as CRM and project tracker
  • Separate paid ticketing
  • Hosted knowledge base subscription
  • Team messaging seats
  • ~10 paid seats growing with headcount

What I built to replace it

  • Kanban ticketing system
  • Internal technical knowledge base
  • Client-facing knowledge base
  • AI meeting → ticket → docs pipeline
  • Public client portal on the same board
  • In-house team messaging (A.V.Y.N.)

The part I'm proudest of

Every company has the same leak: something important gets said in a meeting, everyone nods, and nobody writes it down. Two weeks later the same question gets asked again. I closed that loop.

What happens after a client meeting, with nobody taking notes

A weekly client call is transcribed by Fathom, read by an AI trained on in-house docs, which then opens support tickets and updates the knowledge baseWeekly client callissues raised out loudFathom APItranscript pulled automaticallyAI reads what happenedtrained on in-house docsfinds the real issuesTickets openedrouted to tech support, on the boardKnowledge base updatedthe fix nobody wrote down, capturedZero manual note-taking · zero “I'll write that up later” · zero knowledge left in one person's head

One board, two audiences

Clients wanted visibility. The tech team needed somewhere to talk candidly. Both got what they needed from the same tickets. The portal decided what each side was allowed to see.

What the client saw

  • · Their own tickets on the Kanban board
  • · Real progress as work moved between columns
  • · A comment thread to add detail or chase
  • · A close button, for when it fixed itself

What stayed internal

  • · Technical discussion between me and the tech team
  • · Root-cause notes and half-formed theories
  • · Internal KB articles written for staff, not customers
  • · CEO oversight across both views at once
Full technical breakdown

Same engagement: scaled the company 3 → 13 and built the 10-person CS team.

The compounding part · built at TutorMax

Own every system, and one model can read all of it

This is the argument for in-house that nobody makes up front. A company running on nine vendors has its operational reality split across nine silos and nine terms-of-service. Because I had built the systems, everything the business did landed in one place, and once it was in one place, a model could learn from the whole of it. This was the last thing I built there; here's how it worked.

Every in-house system feeds one model, which answers questions on demand and raises issues on its ownEVERYTHING THE COMPANY DOESMeetingsTeam chatSupport ticketsCallsSales & accountingHR & schedulingIn-house modeltrained on company data only2–3 month learning periodAsked on demandKPIs, revenue, what a milestone neededRaised on its ownpatterns nobody escalated upwardCompany-owned · owner had full access · nothing left infrastructure the business controlled

Asked on demand

Leadership asked in plain language. It logged the question, sifted everything it held, and came back with the answer, and where the question was about a target, what it would actually take to get there.

  • “How many tickets were opened this quarter?”
  • “How many sales did we close, and by whom?”
  • “What's gross income before expenses?”
  • “What do we need to hit the next milestone?”

Raised on its own

The half of the job nobody automates. When teams discussed a problem among themselves that management would never have heard about, the model saw it, and judged whether it mattered yet.

  1. 1

    Mentioned

    Two engineers grumbled about the same thing in a channel nobody senior read.

  2. 2

    Pinned

    The model noticed and held onto it. No action yet: one complaint isn't a signal.

  3. 3

    Recurring

    It came up again. And again. The pattern crossed the threshold.

  4. 4

    Raised

    It surfaced to management, with the history of every time it had been mentioned.

  5. 5

    Checked

    It searched the internal tools for something that already solved the problem.

  6. 6

    Ticketed

    Where nothing did, it opened a ticket. The issue entered the system on its own.

Two to three months of reading first

This isn't a chatbot you switch on. It spent the first two to three months wired in and reading, learning how the company actually talked, what ticket volume normally looked like, which complaints were background noise. Anything useful it said after that was grounded in that company's own history, not a general model's guess about businesses at large.

In-house means it stays yours

The model, the training data and the infrastructure all sat inside the business, and the owner had full access. Nothing about how the company operated was uploaded to a vendor, priced per seat, or held hostage at renewal, which is also the only way a system like this is legally and practically possible in the first place. That remains the case for anything I build.

See me build a neural network from scratch

The live demo implements backpropagation by hand, with no ML library. It's the groundwork this runs on.

“He's not just completing work, he's improving how the work gets done.”
PorfirioChief Operating Officer·FunnelPros.ai / Nonstop Automation
Client endorsement, following a one-month engagement

The full endorsement, unedited

Reproduced exactly as written.

Kenny is an extraordinary individual and a true standout in every sense of the word.

From the moment he steps into a Zoom room, whether it's with clients or colleagues, he makes an unforgettable impression. Kenny combines exceptional performance with genuine charisma, creating a rare blend of talent and personality that elevates every environment he's part of.

He consistently goes above and beyond expectations. Whether it's developing new systems, streamlining operations, or simply executing tasks with excellence, Kenny approaches everything with a high level of ownership and purpose. He's not just completing work, he's improving how the work gets done.

Kenny is deeply professional. He's always punctual, usually early, and willing to stay late when needed. In a time when reliability and commitment are increasingly rare, Kenny embodies both. He's also extremely task-oriented and driven, making him incredibly efficient without ever sacrificing quality.

Most notably, Kenny is a natural-born leader. His presence inspires others, and he brings out the best in those around him. He leads by example, and his positive attitude and high standards create a culture of excellence.

Simply put: if you have the opportunity to add Kenny to your team, take it. He is a game-changer, and any organization would be lucky to have him.

Experience

I've run the departments I build software for

I scale early-stage startups by building the systems they run on. I come in on the customer side (owning retention, onboarding and support), then design and program the internal tooling that lets a lean team operate like a much larger one: custom CRMs, knowledge bases, messaging platforms and automated workflows that replace expensive SaaS subscriptions with software the company owns. I build and lead Customer Success departments from the ground up so founders can stay on strategy and sales, while operations, retention and support run cleanly underneath them.

  1. Head of Client Success

    ·TutorMax.io

    Oct 2025 – May 2026 · Digital marketing · Remote

    Cut tooling spend from $15,000/month to $500 by replacing the stack with software I built, then built the department that runs on it.

    $15k → $500monthly tooling spend3 → 13headcount scaled40%faster onboarding
    • Replaced ClickUp (serving as both CRM and project tracker) and the surrounding paid tools with company-owned software, taking recurring cost from ~$15,000/month to ~$500/month for infrastructure plus GoHighLevel
    • Built a Kanban ticketing system, an internal technical knowledge base and a separate client-facing knowledge base, in JavaScript and light Python with React, TypeScript and Node.js
    • Built an AI layer trained on in-house documentation that pulls weekly client-meeting transcripts through Fathom's API, identifies the issues actually discussed, opens tickets for tech support, and backfills the knowledge base with the fixes and context nobody wrote down at the time
    • Shipped a public client portal onto the same Kanban system: clients tracked their own tickets, commented, and closed them when an issue resolved itself, while internal technical discussion stayed hidden and the CEO retained full oversight
    • Built an in-house neural network wired across every department: meetings, chat, tickets, calls, sales and accounting. Because the systems were all company-owned, one model could learn from the whole operation: it answered C-level questions on demand and independently escalated recurring issues that never reached management, opening tickets where no existing fix applied.
    • Built a team messaging platform in-house, removing another per-seat subscription
    • Scaled the company from 3 to 13 people, personally building out a 10-person Customer Success function: hired, trained and onboarded the whole team and wrote the SOPs they operate by
    • Cut onboarding response times by 40% by documenting and automating the full client-onboarding workflow
    Shipped hereTutorMax Internal PlatformA.V.Y.N.

  2. Customer Success Consultant

    ·FunnelPros.ai / Nonstop Automation

    May 2025 – Jun 2025 · Remote · Contract

    Brought in to fix Customer Success for 80+ clients. Did it in one month.

    80+clients stabilised1 monthengagement
    • Diagnosed and repaired Customer Success failures across 80+ SaaS clients in the construction sector
    • Owned Tier 2 and Tier 3 escalations and resolved the recurring CRM and workflow breakdowns behind them
    • Partnered with engineering on immediate fixes while rebuilding the SOPs, reducing churn and lifting engagement inside a single-month engagement

    “He's not just completing work, he's improving how the work gets done.”

    COO, FunnelPros.ai / Nonstop Automation · read the full endorsement

  3. Chief Operating Officer

    ·Next Wave Services

    Sep 2024 – Sep 2025 · Creative studio · Remote

    Built an operation that ran without me, and stepped back from full-time to advisory as it matured.

    30%faster turnaround0daily oversight needed
    • Ran operations as COO for a fast-paced startup studio, owning production and client delivery while scaling output without losing quality or timelines
    • Designed and implemented AI-driven workflow automations that cut project turnaround time by 30%
    • Built a self-sustaining operational system that ran without daily oversight, which is why the role moved from full-time to part-time to advisory, freeing leadership to focus on growth
    • Led and mentored cross-functional teams across operations, engineering and marketing

  4. Resolutions Manager

    ·B2B Rocket

    Mar 2024 – Aug 2024 · AI-powered SaaS · Remote

    Took dispute losses from roughly 90% to roughly 10% in two months.

    90% → 10%dispute losses8managers led
    • Led a team of 8 Customer Service Managers handling high-priority escalations and disputes for an AI SaaS platform
    • Redesigned the internal resolution workflow, cutting dispute losses from ~90% to ~10% within two months
    • Directed root-cause analysis on recurring technical issues and trained staff on AI-assisted resolution to speed response times

  5. Web Developer · Inventory & Social Media Manager

    ·Silver Ahmet

    Jun 2023 – Nov 2023 · Sterling silver retail

    Built the storefront and the inventory system behind it.

    2×online sales+95%order accuracy
    • Built and launched a fully integrated eCommerce store plus a custom inventory-management system
    • Increased order accuracy by 95% and doubled online sales
    • Ran social campaigns that grew engagement and traffic by 75%
    Shipped hereSilver Ahmet

  6. Data Entry & Payroll Specialist

    ·KMG Consulting

    Mar 2020 – May 2023 · Remote

    Where the EOS and OKR groundwork came from.

    • Managed the firm's payroll system, ensuring accurate processing of wages and benefits
    • Developed and optimised operational systems using OKR and EOS methodologies
    • Led remote teams with performance-tracking and accountability frameworks
    • Provided technical support for VPNs, security protocols and remote-work tooling
Live demos

6 working applications, running on this page

The CRM, POS and Ops Console are the same class of system I build to replace a company's subscriptions. The 3D editor, canvas and neural network are there to show where the ceiling is. All of them run in your browser right now. Click in and use them like real software, because they are.

The whole thing, connected

Ops Console

A working company you can run, or an empty one to put yours into

React 19ZustandDeterministic query engineVitestJSON export
Business systems

Pipeline CRM

Deal pipeline, contacts, forecasting: the whole sales surface

React 19Zustanddnd-kit
Business systems

Retail POS

A till that actually does the money math correctly

React 19ZustandInteger-cent math
Graphics

3D Scene Editor

Build a scene, move things with a gizmo, export the JSON

Three.jsReact Three Fiberdrei
Creative tools

Vector Canvas

Draw, transform, layer, undo, export: a design tool from scratch

React 19SVGCustom transform math
Machine learning

Neural Network Lab

Backpropagation implemented from scratch, training live in front of you

TypeScriptWeb WorkersCanvas 2D
Selected work

4 flagship builds

The largest systems in the catalog. Each one is a product with real users in mind, not a weekend exercise.

All 12 projects
01Game engine / Creator platformIn development

PlayMaker

A no-code 2D-to-3D game engine, built to rival Roblox

The largest project in this catalog: a complete game development engine where you never write code. Physics, destruction, lighting and fluid simulation are all authored visually, and projects scale from 2D up to full 3D on the same platform. Everything is produced in-house through built-in workshops, so a creator never has to leave for third-party software: no Blender, no Photoshop, no external asset pipeline.

ReactTypeScriptThree.jsBabylon.jsClerkAWS S3ExpressDrizzle ORM

The hard parts

  • A real engine, not a tutorial sandbox: physics, destruction, dynamic lighting and fluid simulation, all driven through no-code authoring
  • 2D and 3D on one platform, so a project can grow in scope without being rebuilt somewhere else
  • In-house workshops for assets, logic and scenes. The entire creation pipeline lives inside the product, which is the actual moat
  • Two 3D engines integrated behind one editor, Three.js and Babylon.js, rather than betting the platform on a single renderer
  • Tutorial → Sandbox → Publish → Share, so a complete beginner ends their first session with something playable and shareable by link
  • Clerk-backed auth with S3 asset storage over presigned upload URLs
  • QA run as a written bug report, every issue tracked with root cause, fix and verification status
02Live-streaming SaaSIn development

SuperTurkBot

A Twitch creator platform, its OBS plugin, and its browser extension

A subscription platform for Twitch streamers spanning three separately-distributed products that talk to each other: a web app, a native OBS Studio plugin, and a cross-browser extension. The plugin encodes on the streamer's machine, the platform routes, and the extension personalises per viewer, which is what makes the two headline features possible.

ReactTypeScriptExpressPostgreSQLDrizzle ORMtmi.jsdiscord.jsAnthropic SDKAWS S3WebSocket

The hard parts

  • JustTheMic. The streamer publishes their audio as separate tracks (mic, game, music, guests) and each viewer mixes their own balance in the browser. Turn the game down, mute the music, keep the voice. Nobody else lets the audience control the mix.
  • Live captions, translated per viewer. The OBS plugin encodes captions locally in real time, the platform relays them, and the extension translates into whatever language that particular viewer speaks. A streamer broadcasting in one language reaches an audience in all of them, without doing anything differently.
  • A native plugin, a cloud relay and a browser extension forming one low-latency pipeline: three distribution channels, three release processes, one product
  • 37 numbered development phases with separate public, technical and internal changelogs. The release process is a system, not an afterthought
  • Hand-rolled Postgres migration runner with drift history, so schema changes are reversible and auditable
  • Real-time Twitch IRC ingestion via tmi.js plus a Discord voice bridge, both surviving reconnects and rate limits
  • Encrypted credential storage using @noble/ciphers for third-party OAuth tokens
03Internal comms platformIn development

A.V.Y.N.

In-house team messaging and voice: own it instead of renting seats

Team messaging and voice built to be run in-house. The premise is cost structure: Slack and Teams charge per seat forever, and a company that hosts its own communication layer stops paying that tax and stops handing its internal conversations to a third party. Messaging, voice rooms and auth, deployable on infrastructure the business already controls.

ReactTypeScriptExpressPostgreSQLDrizzle ORMFirebase AuthDaily.coAndroid

The hard parts

  • Built for and deployed at a real company. It was part of the TutorMax stack that took monthly tooling spend from ~$15,000 to ~$500
  • Removes per-seat licensing entirely. The cost of adding the 200th user is the same as the 20th
  • Self-hosted deployment means internal conversations never leave infrastructure the company owns
  • Firebase Auth as the single authentication gate, verified server-side with the Admin SDK
  • Daily.co voice rooms with lifecycle managed server-side, so clients never hold API credentials
  • Ships an Android app alongside the web client, so the team is reachable off a desk without adding a second per-seat bill
  • Originally built on Matrix. The federated protocol was ripped out entirely once it became clear the product didn't need it. Reducing scope beat defending the sunk cost.
04Real-time multiplayer gameIn development

StreamRacer

Multiplayer browser racing where Twitch chat races against paying drivers

A browser-based multiplayer racing game with a Babylon.js + Havok physics client and a Node API server. Twitch viewers join a race by typing !join and drive as AI bots; paying users drive manually over WebSocket. Both race on the same track, in the same simulation, at the same time.

Babylon.jsHavok PhysicsTypeScriptNode.jsWebSocketpnpm workspaces

The hard parts

  • Havok physics simulation running in the browser with a server authoritative over race state
  • Two control paths into one simulation: chat-driven AI agents and live WebSocket input from human drivers
  • pnpm workspace monorepo splitting game client, API server and shared types
  • Custom track and texture pipeline feeding the Babylon scene
Ownership

You own what I build. In writing.

Worth being blunt about, because it is the whole difference between this and buying software. There is no vendor in this arrangement holding anything back.

Terms of the arrangement
  1. 01

    The system is yours, outright

    Not a licence. Not a subscription. Not a seat count. It is designed for your company specifically, and when the engagement ends you keep all of it: the code, the data, the infrastructure and the right to change any of it without asking me.

  2. 02

    Delivered into your GitHub

    You provide a company GitHub organisation and everything lands there. Not my account with an invite for you, yours from day one. If we stop working together tomorrow, nothing needs to be handed over, because it was never anywhere else.

  3. 03

    All code is open to you

    No obfuscation, no black boxes, no parts only I can maintain. Bring on any developer at any point and they can read it, extend it or take it over. Being replaceable is the point: a system that depends on one person is a liability, not an asset.

  4. 04

    Your data never leaves your infrastructure

    Everything runs on servers in your name, under your billing. No third party holds your operational history, prices it per seat, or renegotiates at renewal.

What it runs on, and what it costs
Hetzner Cloud VPSThe server everything runs on~$25/mo
CoolifySelf-hosted deploys, no PaaS billFree, self-hosted
CloudflareDNS, TLS, CDN and DDoS protectionFree tier
GitHubYour org, your repositoriesFree / existing plan
Typical running costabout $25 a month

Put that next to the alternative

The stack above is the same one serving this site. It is boring on purpose: a standard VPS, an open-source deploy tool, a free CDN and a Git host you probably already pay for. Nothing exotic to renew, nothing that scales its price with your headcount.

A company I did this for went from roughly $15,000 a month in subscriptions to roughly $500. Most of that $500 was not even infrastructure.

Open to new work

Tell me where the work is leaking.

The manual step nobody wants to own. The tool you pay for and half-use. The report that's always a week old. Send me the problem. I'll tell you honestly whether I'm the right person for it, and what it would actually take.

Hire me on Upwork

Contracting and payment run through Upwork, so you get their escrow, milestone protection and dispute process on every engagement.

Head of Customer Success / Operations

You need the department and the systems it runs on. I hire and train the team, write the SOPs, and build the tooling underneath, the way I did taking one company from 3 people to 13.

Operations automation build-out

You're paying for nine subscriptions and someone is still copying between them. I replace that with one platform you own: onboarding, ticketing, approvals, reporting, projections. No per-seat bill.

Fractional COO

Own delivery, build the operating system underneath it, then hand it back running. I've done exactly this: full-time to part-time to advisory as the automation matured.

The weird one nobody else wants

A native plugin for a desktop app. An extension operating inside someone else's DOM. Physics in the browser. An ML pipeline stitching four providers together. That's the fun part.