Services
What we do
We take on a few engagements at a time, each with one or two senior engineers doing the work themselves.
Architecture review
- Typical shape
- One to two weeks, one architect, priced in the scope.
- For whom
- Startups whose system is showing strain, or that face a decision (a rewrite, a migration, a scaling step) and want a second opinion before spending on it.
- What you get
- A written assessment of the current system: what is sound, what will break first, and why
- C4-style diagrams of the system as it is and as we propose it
- A prioritised list of fixes with rough effort and the order to do them in
- Architecture Decision Records for the two or three decisions that matter most
- A read-out call with your team
- How it starts
- A 30-minute call, then read access to your repository and infrastructure. We send a one-page scope and the price before starting.
- Not a fit when
- You want a rubber stamp on a decision already made.
MVP build
- Typical shape
- Eight to fourteen weeks, one or two engineers, invoiced against the phased plan.
- For whom
- Founders with a validated problem who need a working first version, built so a future team can keep working on it.
- What you get
- A scoped plan: epics and tasks with acceptance criteria, in your tracker or ours
- The product running on AWS, with infrastructure defined in code (AWS CDK) and a CI/CD pipeline
- Automated tests on the paths that matter, and a README that gets a new engineer running in an hour
- A written update every week, and demos at the cadence we agree in the proposal
- Handover built in: repository, cloud accounts and documentation in your name from day one
- How it starts
- Two calls, one on the product and one on the scope. Then a written proposal with a phased plan, a price, and the things we are deliberately leaving out.
- Not a fit when
- The scope is undefined and you want a fixed price on it anyway, or you need a large team by next month.
Embedded senior engineering
- Typical shape
- Three to six months, one engineer, part-time, or full-time for a defined stretch when we can staff it. Invoiced monthly.
- For whom
- Teams that need a senior TypeScript, Node.js or AWS engineer for a defined stretch: a migration, a hard feature, or cover while you hire.
- What you get
- Shipped work in your repository, following your process and your reviews
- Code review and pairing with your engineers
- Decision records for the calls made on our watch, so context doesn't leave when we do
- A short written update every week
- How it starts
- A call about the work and your stack, then a short paid trial at the agreed rate, with its length set in the proposal. Either side can stop at the end of it.
- Not a fit when
- The work is outside our stack (Python, Java, .NET, native mobile) or needs someone on-site.
AI features on existing products
- Typical shape
- Three to six weeks, one engineer; the spike is priced separately from the build.
- For whom
- Product teams that want document extraction, classification or an assistant inside their product, without hiring for machine learning.
- What you get
- A one-week feasibility spike on your real data, with accuracy numbers and cost per document
- A production pipeline on hosted model and document-extraction APIs, with confidence scoring and a review step for low-confidence cases
- An evaluation set and regression tests, so a prompt change doesn't silently break results
- Cost and latency monitoring from day one
- How it starts
- Send us a sample of the documents or the workflow. We run the spike first; you continue only if the numbers hold up.
- Not a fit when
- You need a custom-trained model, or your volume makes API pricing the wrong economics.
Contact
Tell us what you're building
Remote, from Montevideo, Uruguay (UTC−3, no daylight saving): a full working-day overlap with US Eastern, and mornings with US Pacific.