Engineering
From first customer
to fiftieth without a rewrite
Product-market fit tends to break the architecture that found it. We take startups from their first paying customers through multi-tenancy, security questionnaires and the first in-house engineering hires, without pausing feature delivery to do any of it.
Overview
The eighteen months after launch, when scale, hiring and enterprise buyers all arrive.
The systems that carry a startup to its first hundred customers are usually correct for that stage and wrong for the next. A shared database with a tenant column, onboarding done by hand, one person who knows how the deploy works: all defensible at ten customers, all dangerous at three hundred. The skill is sequencing the fixes so none of them needs a stop-the-world release.
We operate as an embedded product team, typically a tech lead, three to five engineers and a product designer who sit in your planning sessions. Roadmap authority stays with you. What we bring is the set of things most seed-stage teams have not built before: tenant isolation, billing that survives a mid-cycle plan change, audit logging and job infrastructure that does not silently drop work.
The engagement is designed to shrink. As your own hires land, we transfer domains one at a time against an agreed schedule and reduce our headcount accordingly. An arrangement that only ends when someone questions the invoice was structured badly from the start, and we would rather say that up front.
- 8 to 240
- Tenants supported after the isolation rework at Halden Group
- 99.95%
- Measured availability across fourteen months of the engagement
- 6 engineers
- Hired and onboarded into the codebase before our team stepped down
Capabilities
What this covers
Six areas we staff properly. If your problem sits outside them, the honest note at the foot of this page says so.
Multi-tenancy and isolation
Row-level security or schema separation chosen against your compliance obligations and query patterns, migrated behind a feature flag while the single-tenant path keeps serving existing customers.
Billing that survives reality
Plan changes mid-cycle, proration, usage metering, dunning and failed-payment recovery built on Stripe Billing, with reconciliation against your ledger so finance is not correcting revenue by hand each month.
Enterprise readiness
SAML and SCIM through WorkOS, granular role permissions, immutable audit logs and the evidence collection that makes a SOC 2 Type II observation window an administrative task rather than a fire drill.
Scaling the data layer
Query analysis before hardware, indexes and partitioning where the traffic is, read replicas and connection pooling with PgBouncer, and background work moved off the request path into durable queues.
Delivery infrastructure
Staging that mirrors production, trunk-based development with feature flags, deployment several times a day, and alerting tuned so an on-call engineer is woken only by things that are genuinely broken.
Hiring and knowledge transfer
Technical interview design, hands-on assessment of candidates, and a domain-by-domain handover schedule so each new engineer inherits a documented area rather than a codebase-shaped mystery.
Deliverables
What you get
- Embedded squad with named engineers, a tech lead and a designer in your planning
- Multi-tenancy migration plan and executed rollout with per-tenant isolation verified
- Billing system reconciled against the ledger, covering proration and failed payments
- SSO, SCIM, audit logging and a security questionnaire response pack for enterprise buyers
- Runbooks, alert policies and an on-call rotation your own engineers can take over
- Quarterly handover schedule mapping each domain to the in-house owner taking it on
Stack
What we build it with
- TypeScript
- Next.js
- NestJS
- PostgreSQL
- Prisma
- Redis
- BullMQ
- Terraform
- AWS
- Stripe Billing
- WorkOS
- Datadog
Process
How the engagement runs
Two-week increments against a written definition of done. You can stop at any increment boundary and keep everything built so far.
Read the system and the pipeline
Three weeks reviewing architecture, incident history and the deals your sales team is losing on technical grounds, which usually reveals the real priority order.
Sequence the structural work
Multi-tenancy, billing and security work ordered against contract dates and customer growth, with each item sized so it fits alongside continued feature delivery.
Run two tracks in parallel
A feature track and a foundations track sharing one backlog and one standup, with an agreed split, commonly seventy-thirty, that is reviewed every month.
Prove it under load
Load and failover testing at projected twelve-month volumes, plus a tenant isolation review, before enterprise contracts commit you to numbers in a service agreement.
Shrink deliberately
Domains handed to your engineers as they join, our headcount reduced on the agreed schedule, and a final review of what remains undocumented before we exit.
When this is the wrong engagement
If you are pre-revenue and still testing whether anyone wants the product, this is premature investment, and a scoped MVP is the right starting point until customers are paying and the seams are visible.
FAQ
Questions we get asked
- How is this different from hiring an agency?
A project agency delivers a defined scope and leaves. We join your planning, hold on-call responsibility alongside your team and are measured on product outcomes rather than delivered tickets. The engagement is also written to reduce as you hire, which a scope-based contract has no reason to do.
- Who owns the roadmap?
You do, without exception. Our tech lead brings sequencing arguments, cost estimates and the occasional strong objection, but prioritisation sits with your founders or product lead. If we think a decision is wrong we say so in writing once, then build it properly.
- Do you take equity instead of fees?
No. Equity arrangements create the wrong incentives on both sides and complicate your cap table before a priced round. We work on monthly fees with a thirty-day notice period, so ending the engagement is straightforward if it is not working.
- Can you support technical due diligence for a raise?
Yes. We prepare architecture documentation, security posture summaries, dependency and licence audits, and the incident history investors ask for. We also brief your CTO or founder on the questions that get asked, since the answers matter more than the documents.
Related
More in Engineering
Engineering
Blockchain Solutions
Permissioned ledgers, tokenised assets and independently audited smart contracts.
Engineering
IoT Development
Firmware, connectivity and telemetry pipelines for devices you cannot easily reach.
Engineering
Legacy Modernisation
COBOL, Delphi, WebForms and Oracle Forms moved off, without a big-bang cutover.