taking on new builds

capabilities

Systems we design, build, and run.

Web platforms, native apps, AI agents, and the payments, data, and infrastructure underneath them. Schema to pixel, with tests on every merge.

Scope a build See the architecture

~/bigdevelop — zsh

online

# what one team ships, end to end$bd capabilities --layersclients    web · ios · admin · voiceservices   api · payments · agents · jobsdata       postgres · queues · storageops        ci · previews · alerts · backups✓10 disciplines · 1 accountable team→ ready to scope your build

disciplines

Ten disciplines, one accountable team.

Most products need several of these at once. We cover all of them in-house, so the database, the app, and the payment flow are designed by the same people.

platforms
Full-stack web platforms and two-sided marketplaces
Multi-tenant SaaS, marketplaces that match supply with demand, and the admin consoles that run them. One team owns the schema, the API, and every screen.
- Role-based auth with row-level security
- Realtime updates, search, and notifications
- Onboarding, verification, and review flows
- Internal admin and support tooling
```
-- tenants only ever see their own rowsalter table orders enable row level security;create policy tenant_read on orders  for select using (    org_id = auth.jwt() ->> 'org_id'  );
```
next.jstypescriptpostgressupabase
mobile
Native iOS and mobile apps
SwiftUI apps with offline-first sync, widgets, and push, or cross-platform builds when one codebase is the right call.
- Offline sync with conflict handling
- Widgets and push
- App Store release pipeline
swiftuiflutterreact native
agents
AI voice agents and AI employees
Phone agents that answer, qualify, book, and hand off to a human. Tool-calling agents that operate your internal systems with an audit trail.
- Tool calling against your real APIs
- Guardrails and human handoff
- Transcripts and eval sets
claudeopenaitwilioplaywright
payments
Payments and billing
Card and ACH, subscriptions, invoices, and marketplace payouts through connected accounts, with idempotent webhooks and daily reconciliation.
- Checkout, subscriptions, and dunning
- Split payments and payouts
- Ledger and reconciliation
stripestripe connectach
integrations
Automation and third-party integrations
CRMs, accounting, calendars, messaging, and legacy tools wired together with queued jobs, retries, and alerts when something upstream breaks.
- Webhook ingestion with retries
- Scheduled jobs and backfills
- Two-way sync
restgraphqlwebhookscron
data
Data pipelines and dashboards
Pull from every system you use, normalize it into Postgres, and put the numbers that matter on one screen, or in your inbox every morning.
- ETL from third-party APIs
- KPI dashboards
- Scheduled reports
sqlpythonpostgres
hardware
Hardware and IoT integrations
GPS trackers, Bluetooth devices, and always-on displays connected to real backends, with telemetry you can query.
- BLE pairing and device control
- Location and telemetry ingest
- Kiosk and signage modes
core bluetoothgpsmqtt
infra
Infrastructure, DevOps, and CI
Preview environments on every pull request, migrations in CI, secrets in a manager, and each project isolated in its own accounts.
- Preview deploys per branch
- Migrations as code
- Per-project credentials
github actionsvercelawsdocker
testing
Automated testing and QA
Unit, integration, and end-to-end suites that run before anything merges, plus contract tests on the webhooks that move money.
- 80% coverage floor
- E2E on critical flows
- Regression suite per release
vitestplaywrightxctest
security
Security built in, not bolted on
Least-privilege keys, row-level security on every table, secrets never in source, rate limits on public endpoints, and an audit log for anything that touches money or personal data.
- Row-level security policies
- Secrets management and rotation
- Rate limiting and input validation
- Audit trails
rlsoauthjwtowasp top 10

architecture

Follow a request through the whole system.

Most of our builds share one shape: typed clients, a thin edge, domain services, Postgres, and the outside providers they depend on. Pick a flow to see the path it takes.

clients
- 1web app, hop 1
- ios app, not on this path
- admin console, not on this path
- phone caller, not on this path
edge
- 2cdn + route handlers, hop 2
- 3auth / jwt, hop 3
services
- 4domain api, hop 4
- 5payments svc, hop 5
- agent runtime, not on this path
- job workers, not on this path
- 7webhook ingest, hop 7
data
- 8postgres + rls, hop 8
- job queue, not on this path
- object storage, not on this path
external
- 6stripe, hop 6
- voice / sms, not on this path
- llm provider, not on this path
- third-party apis, not on this path
- apns, not on this path

observability on every layer: structured logs · traces · alerts

trace

A buyer pays, the platform takes a fee, and the seller is paid out, with no double charges if a webhook retries.

1
web app
buyer confirms order
2
cdn + route handlers
request hits a route handler
3
auth / jwt
session verified, org resolved
4
domain api
order created in a transaction
5
payments svc
payment intent with idempotency key
6
stripe
charge + transfer to seller account
7
webhook ingest
signed event verified, deduped
8
postgres + rls
ledger row written, order marked paid

stack

Tools picked for the next five years.

We choose mature, well-documented technology your next hire will already know. No exotic frameworks, no lock-in to us.

languagestyped wherever the platform allows: typescriptswiftpythonsqldart
webserver-rendered, fast on slow phones: next.jsreacttailwind cssframer motion
mobilenative first, cross-platform when it fits: swiftuiuikitwidgetkitflutterreact native
backendboring, proven, well-indexed: node.jspostgressupabaseedge functionsrest + webhooks
aimodels behind tools, evals, and guardrails: claudeopenaitwilio voiceelevenlabsplaywright
paymentscard, bank, and marketplace money movement: stripestripe connectachplaid
infrareproducible, isolated per project: vercelawsgithub actionsdocker
qualityruns before anything merges: vitestjestplaywrightxctesteslint

engineering process

Shipped like software that has to stay up.

Speed comes from discipline, not shortcuts. The same checks run on a two-week build and a two-year platform.

Tests before merge
Unit and integration tests on business logic, end-to-end tests on the flows that make money. CI blocks the merge if anything fails.
Every change reviewed
Small pull requests, a written description, and a second engineer on anything touching auth, payments, or data migrations.
Observability from day one
Structured logs, error tracking, and traces wired in before launch, so production issues show up with context attached.
Uptime you hear about first
Health checks and alerts on the endpoints that matter. If something breaks at 2 a.m., we know before your customers do.
Security reviewed, not assumed
Secrets in a manager, least-privilege keys, validated inputs, and a security pass before anything handling money or personal data ships.

~/ci — pull request #214

passing

# runs on every pull request$git push origin feat/seller-payoutsci: typecheck → lint → unit → integration → e2e✓tsc --noEmit            0 errors✓unit + integration      412 passed✓playwright e2e           38 passed✓coverage                 86.4% (floor 80%)→ preview deployed · review requested

default budgets on every build

test coverage: 80%floor
uptime: 99.9%target
p95 api latency: 300msbudget
deploys: everymerge

engagements

How a build runs, week by week.

A small senior team, one point of contact, and working software from the first week.

01scope
week 0
Scope
A working session on the problem, the users, and the systems you already run. You get a written scope, an architecture sketch, and a fixed price or a monthly plan.
02architecture
week 1
Architecture
Data model, integrations, and risks first. Accounts are created in your name, so the code, cloud, and payment provider are yours from the first commit.
03build in increments
weekly
Build in increments
A preview link every week with working software on it, not slides. You test real flows and steer while changes are still cheap.
04launch and run
ongoing
Launch and run
Production cut-over with monitoring, backups, and a runbook. Stay on for support and new features, or take it in-house with full handover.

Three ways to work with us

what you own at the end

Source code in your repository
Cloud, database, and payment accounts in your name
Docs, runbooks, and environment setup
No per-seat licenses to us

next step

Have a system in mind? Let's scope it.

Tell us what it needs to do and what it has to connect to. We'll come back with an architecture, a timeline, and a price.

Scope a build See services

reply within one business day

Systems we design, build, and run.

Ten disciplines, one accountable team.

Full-stack web platforms and two-sided marketplaces

Native iOS and mobile apps

AI voice agents and AI employees

Payments and billing

Automation and third-party integrations

Data pipelines and dashboards

Hardware and IoT integrations

Infrastructure, DevOps, and CI

Automated testing and QA

Security built in, not bolted on

Follow a request through the whole system.

Tools picked for the next five years.

Shipped like software that has to stay up.

Tests before merge

Every change reviewed

Observability from day one

Uptime you hear about first

Security reviewed, not assumed

How a build runs, week by week.

Scope

Architecture

Build in increments

Launch and run

Three ways to work with us

what you own at the end

Have a system in mind? Let's scope it.