Juan Manuel GarcíaBuenos Aires, Argentina · remote worldwide
Twenty-one products in production.
Python & full-stack engineer. I build the product, the infrastructure it runs on, and
answer the phone when it breaks: a network of Shopify apps used by 10,000+ merchants,
DeployCloud, a managed PaaS you can drive from a chat, and Crawley, a Python
crawling framework the community starred ~200 times.
A managed platform-as-a-service you can drive from a chat. git push — or ask Claude, Cursor
or any MCP client — and you get a live HTTPS app in any language: zero-downtime deploys, preview
environments, managed add-ons (Postgres, MySQL, Redis, pgvector), cron jobs, autoscaling and
one-click rollback. Built and operated solo, from the build queue to the TLS certificates.
A network of Shopify apps for reporting, tagging, automation, upsells, SEO and support — used by
10,000+ merchants, including brands like ANINE BING. Django and Python behind the scenes, with
worker fleets chewing through webhooks around the clock.
Custom reports over orders, products, customers and inventory, exported to CSV, Excel or PDF and scheduled by email. It ingests tens of thousands of webhooks an hour across the merchant base, so most of the work is queueing, back-pressure and making a report that spans years come back in seconds.
Tags orders, products and customers automatically from rules a merchant writes in plain language — plus bulk and AI tagging over an entire catalogue. The rule engine runs on its own worker fleet, because a single bulk run can touch every record a store has.
A branded tracking page on the merchant’s own domain, fed by 1,000+ carriers. Tens of thousands of shipments are re-polled every day, with per-carrier scrapers and a scheduler that has to tell a parcel that is moving from one that will never answer again.
Live chat for a storefront through the apps shoppers already have open: WhatsApp, Messenger and the rest. The widget ships as a theme app extension, so it loads with the storefront instead of blocking it.
Scans a store and fixes what it finds: page titles, meta descriptions, image alt text and broken links. The model proposes, the merchant approves, and every change is written back through the Shopify API with a way to undo it.
A “try it on” button on the product page: the shopper uploads a photo and sees themselves wearing the garment. Image generation on a queue, with the photos kept on a volume the merchant can clear.
Fills and maintains metafields across a whole store from rules instead of by hand — the structured data themes, feeds and marketplaces read. Dozens of RQ workers do the filling, because “apply to every product” means every product.
DjangoRedis / RQShopify API
+
The rest of the catalogue apps.shopify.com/partners/dev-cloud
Same stack, same on-call
Twelve more apps on the same Django + Redis backbone — automations, metafields, upsells, discounts,
alerts, social login, AI publishing and storefront widgets. All of them mine, from the first migration
to the pager.
Kanban boards where every card opens like a Notion page. Multi-user SaaS with real-time
collaboration, an API, and the same experience from desktop to phone.
Booking for service providers: publish your hours once and clients take the slot that suits them,
with payments at checkout, business emails, calendar sync and a reschedule link — so nobody has to
trade messages to find a time. Also shipping as a Shopify app.
The astronomy toolkit HOPS
turned into a web app: plan exoplanet observations and fit transit light curves in the browser,
no local install, with the original scientific engine doing the maths.
Live chat on any site plus a shared inbox where a team answers together: canned replies, AI-assisted drafts and a help centre. Sold on its own and as a Shopify app, on the same codebase.
Hosted error tracking and performance monitoring: full stack traces, releases, cron monitors and session replays. It has to stay up precisely when the app it watches does not, and swallow a burst of ten thousand identical errors without becoming the outage.
Turns any website into structured data: you click the fields you want and get clean JSON, CSV or Markdown through an API — with JavaScript rendering, proxy rotation and scheduled crawls. Headless browsers on a queue, which is a fleet-management problem more than a parsing one.
Support email and documentation in one workspace: a shared inbox where a team answers without stepping on each other, a knowledge base customers can search, and reporting on what actually gets answered.
A full online store for merchants who do not want a per-sale commission: catalogue, checkout with real payments, shipping, discounts and a storefront they can restyle. The whole stack behind a sale, run as a product rather than a plugin.
PythonPaymentsMulti-tenant
Open source
On GitHub since 2010
Libraries I published because I needed them first. Star counts as of September 2026.
Plus 66 more: a pure-Python Scrabble engine, a Qt web browser, a NSIS installer generator, a music player, Django routing and deployment plugins… → github.com/jmg
Claude & OpenAI APIsMCP serversAgent toolingpgvectorWeb scrapingPlaywright
Path
Seventeen years, short version
2015 — now
Founder & principal engineer · DevCloud Software
Building and running the Shopify app network and DeployCloud: product, code, infrastructure and support.
2015 — now
Verified Expert in Engineering · Toptal
Member since March 2015, in the network's top 3% of talent.
2011 — 2015
Python web developer · Deviget
2010 — 2015
Freelance web developer · Python, Ruby, PHP
2008 — 2011
Web & C# developer · MSA, Ingematica, TNS Gallup
2008 — 2015
Systems Engineering · UTN, Buenos Aires
Let's talk
Got something hard to build — or something broken to rescue?
I take on back-end, full-stack and DevOps work, Shopify apps, scraping at scale and AI features
that have to survive real traffic. Fastest way in: my Toptal profile.