Platform rebuild
Fintech series-B, ~80 engineers
A 5-year-old SPA replaced with a server-rendered Rails app in 4 months.
- ✓ p95 page load: 2.4s → 280ms
- ✓ Deploy: 22m → 4m
- ✓ 23k LoC removed, net
What we do
Three ways to work with us. All staffed by senior engineers, all opinionated.
01 / Service
Custom development
You bring the problem; we ship the platform.
Polished web platforms and developer tools, built end-to-end by a small senior team. We design, implement, and operate.
- → Greenfield products
- → Rebuilds & migrations
- → Internal platforms
Read more →
02 / Service
Engineering consulting
A senior second opinion on the calls that matter.
Architecture reviews, stack audits, and the difficult tradeoffs you don't want to make alone. We leave you with a written record, not a slide deck.
- → Architecture review
- → Stack & dependency audit
- → Hiring & team structure
Read more →
03 / Service
Embedded engineers
A Upcast engineer shipping inside your team.
Senior engineers on retainer, integrated with your standups, your repo, and your pager. We pick up the work nobody else wants and make it boring.
- → 3-month minimum
- → Part- or full-time
- → Async-friendly
Read more →
How we think
Bold over trendy.
Boring over clever.
These are the calls we make when nobody is watching. They show up in every codebase we touch.
- 01 A working monolith before microservices.
Most teams reach for distributed systems before they have a distributed problem. We start with the simplest topology that fits, and split only when the load tells us to.
- 02 HTMX and server-rendered before another SPA.
If a request-response page handles the workflow, that's the design. Heavy clients earn their place with a measurable interaction need — not because it's the default.
- 03 Postgres and a queue before a bespoke pipeline.
Boring data infrastructure carries you further than any team admits. We pick proven primitives, write them down, and only graduate when there's a real reason to.
- 04 A polished UI on a small surface before a flat scope creep.
Users feel the seams more than the feature count. We'd rather ship five flows that delight than fifteen that almost work.
- 05 Code we can delete before code we can extend.
Optionality is a tax. We write the code our future selves can rip out cleanly when the requirements change — because they will.
- 06 Tests on the fault lines before coverage targets.
100% coverage is a vanity metric. We test the things that, when they break, wake somebody up — and skip the rest with a clear conscience.
The first deliverable
A stack audit you can actually action.
Every engagement opens with a written audit of your codebase, dependencies, build pipeline, and deploy story. No slides. Findings ranked by what wakes someone up, not what scores well in a checklist.
- Format
- Markdown report, ~10 pages. Yours to keep.
- Turnaround
- 5 business days from repo access.
- Cost
- Fixed fee. Credited toward an engagement if you decide to.
Live
Ask a Upcast engineer.
Have a stack question? Get an opinion grounded in how we work — short, direct, and with the tradeoff that would change our mind.
Selected work
Three engagements, anonymized for now.
Stack audit + plan
Logistics scale-up, 12 services
Cut three of twelve services. Kept Postgres. Stayed on the same team size.
- ✓ Infra spend −38%
- ✓ On-call pages −74%
- ✓ New-hire ramp 4w → 1w
Embedded engineering
B2B SaaS, founder-led
A Upcast engineer ran the platform team for two quarters while they hired.
- ✓ SLO 99.5 → 99.95
- ✓ Shipped two product lines
- ✓ Hired 4 backfills
Get in touch
Have a project in mind, or a stack you'd like a second opinion on?
We reply to every inbound within one business day. If we're a good fit for the work, we'll schedule a 30-minute call. If we're not, we'll point you somewhere we trust.