Senior Rails & Go engineer
I build and scale multi-tenant SaaS.
Seven years shipping production SaaS and fintech in Rails, plus Go for the performance-critical parts. Right now I own the backend — billing, scheduling, document generation — for a multi-tenant platform serving 1,000+ contractors, and the AWS it runs on.
/ work
A B2B bug-capture tool: one link records a customer's screen, console and network so support can replay the exact repro instead of asking them to explain it. A browser extension and TypeScript recorder up front, a Rails 8 backend, and a Go service streaming the uploads over WebSockets.
A one-click deploy platform for Rails apps on AWS — a Rails orchestrator driving a fleet of Go agents over gRPC. Per-server routing, Docker builds, deploy logs streaming live.
A self-hosted product-analytics platform — event tracking, funnels, retention, anomaly detection and a live dashboard, with zero-dependency SDKs for Ruby, JavaScript, Python and Node. Rails 8, PostgreSQL, Solid Queue, Hotwire.
Roofing and construction estimation — mark measurements up on a blueprint canvas, roll costs with waste, labour and Canadian GST/QST, export print-ready reports. My exact industry at C-CUBE, built solo. Vue 3, TypeScript, Pinia.
An offline-first audiobook study tool, shipped solo to the App Store. Highlight a passage mid-playback, note against it, and review a whole book from one library — with real M4B chapter parsing and everything kept on-device. No accounts, no servers; localized into 10 languages. A web version is now in the works.
/ experience
Lead engineer on a multi-tenant SaaS for 1,000+ roofing contractors — I own the backend for billing, scheduling and document generation. Cut median query time 40% with indexing and query refactoring, built the Sidekiq + AWS SQS jobs running thousands of PDF generations, email syncs and billing events a month, and brought Go in for the CLI tooling that handles tenant data ops, backfills and migrations. I own the AWS side (EC2/RDS/S3), the CI/CD, and the Prometheus/Grafana stack.
Bookkeeping and tax-compliance features for a fintech SaaS serving 10,000+ Australian freelancers. Integrated FreshBooks and Stripe for real-time transaction sync under strict financial-accuracy requirements, and automated invoicing and tax calculation — cutting users' manual bookkeeping by an estimated 60%.
Built ProCRM from zero — a CRM for multinational call-center operations across Tangier and London. Designed a PostgreSQL schema for millions of customer records under heavy write load, and shipped the REST APIs and real-time dashboards management used for cross-office KPIs.
Marketplace features for an agritourism platform across France — booking flows, Stripe payments, listing management. Tuned queries and caching to keep response times down under load.
Shipped 15+ client sites and kept production CMS deployments running for a varied client portfolio. Where I learned to hit someone else's deadline.
App Academy (full-stack web) · OFPPT (computer science)
/ stack
/ How I work
Most outages I've dealt with weren't interesting — a missing index, or a webhook that ran twice on a retry. I build for both before they happen.
/ How I think about it
I'd rather ship something boring that works than something clever I have to babysit. New tools have to earn their place — I've maintained enough of other people's excitement.
Legacy code isn't the enemy. It's running the business, and it usually works for reasons that aren't obvious from reading it once. I read it before I change it, and I change as little as I can get away with.
The goal is a system the next person can understand without me in the room.
Fewer moving parts than the problem seems to need.
/ Things that broke
On a retry, the financial integration replayed a webhook — the kind of thing that quietly throws off reconciliation until the numbers stop matching.
FixIdempotency keys, replay-safe handlers, reconciliation checks. Now an outage just delays the sync instead of corrupting data.
Wrong indexes, and query shapes whose real cost only showed once every tenant's data sat behind the filter.
FixReworked the indexes and the queries — about 40% off median query time. Not clever, just necessary.
Tenant data ops and migrations were one-off Rake scripts — fine on a laptop, painful on production-sized data.
FixMoved them into Go CLI tooling. Runtime on the big batch jobs dropped enough that we stopped scheduling the business around them.
/ Approach
I'm comfortable owning a feature from the database it lives in to the AWS it runs on. It usually goes in three passes.
Design the Postgres for the access pattern and the multi-tenant filter, and index for it before it's a problem. That's where the 40% query win at C-CUBE came from — not cleverness, just shaping the data for how it's actually read.
The async work goes on Sidekiq + SQS — thousands of PDF generations, syncs and billing events a month. Financial integrations get built idempotent and reconciled from the start, so a retry or a partner outage never corrupts the numbers.
It runs on AWS I own — EC2, RDS, S3 — shipped through GitHub Actions, watched through Prometheus and Grafana. So when it moves, someone can actually see it.
Schema to dashboards — one person who can own the whole column.
/ Case studies
Multi-tenant quota enforcement, Stripe billing idempotency, and a reconciliation trail for every generated document.
Multi-tenant data isolation and a hot path that stays under 200ms when a single deploy floods it with thousands of errors — deterministic fingerprinting to group them.
Zero-data-loss time tracking for technicians with no signal — an ACID offline queue, GPS geofencing, and deterministic conflict resolution on sync.
Astronomical prayer-time math on a Garmin watch — fully offline, power-aware GPS, five calculation methods, implemented from first principles with no external libraries.
Small things I use every day, so I built them properly.
