MedViz
A medical data platform that pulls lab results out of unstructured reports through LLM pipelines and standardises them into something you can actually analyse. Started because reading your own bloodwork should not require a translator.
the only KPI that matters is your bottom line — the rest are vanity metrics
Remu Apparel has been my best teacher. It's mine, so every hat is mine too — the product, the store, the media budget, the artwork, the mistakes. You read a business differently once you've had to wear all of them, and once the money on the line has been your own.
Since then I've done that work inside other people's businesses: growth consulting at Apiece Apart, the AI transformation at Confiteca, growth at Part & Sum, development and ecommerce optimisation at ROAM. I built a contribution-margin model that shows me how an ecommerce brand is really performing rather than what the dashboards would like me to believe — I read data the way I read anything else, as a story, and that's usually where the turnaround is hiding. I don't claim to know everything. I'm just curious enough to go and figure it out — and technical enough to build it once I have.
hover it — the ink lifts off
People call me when the problem won't stay in its lane — when the growth question turns out to be a merchandising question, when the site problem turns out to be a tracking problem, when the AI project turns out to be an operations problem. Those are my favourite ones. They're where knowing a bit of everything stops being a party trick and starts being useful.
My job is to help a brand tell its story properly and still make money — and those two things live in different parts of the business, which is why most companies hire two different people and then spend six months translating. Most consultants own one part. I've worked in all eight. That's a big claim, so I built the nerdy side to let you check it: every engagement plotted against the parts it actually touched, nothing rounded up.
Nights and weekends, mostly. I get an itch about a problem, build the thing, then find out other people had the same itch. It's also where most of what I know about applied AI actually came from — you learn more shipping one real product than reading a hundred threads about it.
A medical data platform that pulls lab results out of unstructured reports through LLM pipelines and standardises them into something you can actually analyse. Started because reading your own bloodwork should not require a translator.
GPT-powered prompt packs for ecommerce founders across acquisition, email, operations and finance. Really it's the set of playbooks I kept rewriting from scratch for every client, finally written down once.
A to-do app that never shows you a list. Every capture gets one of four verdicts once a day — Do (capped at three, because the cap is the discipline), Delegate (it drafts the handoff for you), Later (only with a when-trigger, which is what lets your brain release it), Delete (after answering what actually happens if this never gets done). Then one screen, one task. Flag something as dread rather than difficult and it asks you to predict how bad it'll be, then rate the reality — the weekly audit becomes your own proof that the fear is bigger than the thing. No projects, tags, priorities or red badges: those are how to-do apps manufacture anxiety and get abandoned.
The AI co-pilot for connection — a voice-first mystery game for road trips where the AI plays Dungeon Master and the passengers solve a whodunit together. It holds two states at once: the truth only it knows, and what the car has actually discovered. Every behaviour is tuned to make people look at each other instead of their screens — naming passengers, balancing who speaks, celebrating when someone builds on someone else's idea, and scaling the crime to the youngest person in the car.
Uses NLP to read live product size charts across PDPs and recommend a size — a small fix for the single biggest reason apparel comes back.
I draw most days — from life and from reference. It has nothing to do with anyone's P&L and everything to do with why the work comes out looking like something instead of looking like a template. It's also the part I'd keep doing if nobody paid me.
click any of them to open it →
The half that makes people want the thing, and the half that decides whether wanting it ever turns into profit. I've spent eight years refusing to specialise in one, because a brand that only has the first one is a nice-looking business that loses money.
Identity and campaign work, photography and print, the visual system behind Remu Apparel. I can hold a brand's aesthetic standard in the room and then go and execute it myself, which is the only reliable way I've found to stop good strategy getting watered down on its way to the artwork.
Art direction · Identity · Campaign · Print · Photography
Contribution margin and P&L, paid and lifecycle performance, Shopify theme development, product customizers, GA4 and Ads instrumentation, and applied AI — building it, knowing where it actually belongs in a business, and using it well day to day. The unglamorous layer that decides whether the beautiful thing sells, and the one most creative people quietly avoid.
Growth · Ecommerce ops · Web development · AI & automation
I'm taking on new consulting work right now — ecommerce and growth, AI builds, or the messy ones that are both. Send me a paragraph about the business and the thing that isn't moving. No form, no discovery funnel. And if it turns into nothing, I'll still tell you what I'd do about it.
josemchiriboga@gmail.comlinkedin.com/in/josemchiriboga
English (fluent) · Spanish (native)
BS Computer Information Systems, minor in Entrepreneurship — College of Charleston, 2019
Florence University of the Arts, 2016
Student Entrepreneur of the Year, 2019 · Outstanding Entrepreneurship Student, 2019 · 1st place, New Venture Modeling & Business Plan pitch competitions, 2017
Claude & Claude Code · MCP servers & tool use · agent design · LLM pipelines · RAG · prompt engineering · OpenAI · Higgsfield · generative image, video & 3D · Cursor · Python · JavaScript · HTML/CSS · Shopify · WebGL
Meta · Google · TikTok · LinkedIn Ads · Klaviyo · GA4 · Looker Studio · contribution-margin modelling · forecasting · Photoshop · Illustrator · InDesign · Premiere · Figma
Same content, linear. Every engagement, the stations of the commerce loop it touched, and what the work actually was.
You're on the nerdy side. The eight parts of an ecommerce business sit on the ring, and every engagement inside it is wired to the parts it actually touched — nothing rounded up. Drag to move, scroll to zoom, click a node to open it — or hit Run the loop and let it drive.