Engineering for software that runs the business. Year after year.

Production AI engineering.Claude, MCP, eval suites, HITL gates, retrieval design, prompt versioning, audit logging. The substrate that makes a model survive past month three inside a real operation. If your last shipped AI feature broke when the prompt changed, this isn’t the practice for you.

Operational platforms.NestJS, Postgres, Next.js, integrations into SAP, Salesforce, NetSuite, Procore, ServiceNow. Software a dispatcher signs into at 6 a.m. and stops the business when it’s down. Boring is the feature.

Mobile field tools. React Native first, native iOS (Swift) or Android (Kotlin) where the platform demands it. Offline-first sync, durable queues, conflict resolution. Driver apps, inspection apps, two-sided marketplaces.

We hire mid-level engineers occasionally, only into embedded retainers where the runway lets them grow inside production code. We don’t hire junior. The first engagement carries too much weight for someone still finding the fundamentals.

Not portfolio sites. Not internal tools that died when the contract ended. We hire engineers who have shipped systems that survived the year. Bonus: the story of the year-two refactor that came out of operating it. We learn the most about a candidate from that conversation.

Comfortable taking a workflow from problem to production. That includes the unglamorous middle: retries, error budgets, observability, eval suites, on-call handoffs. If the interesting work to you stops at the happy path, we’re not the right shop.

Honest about what you’ve shipped versus what you’ve watched others ship. The interview will reveal it either way. Save the embarrassment.

Phoenix office in Tempe and a Los Angeles office on Mira Loma. Hybrid for senior engineers in either metro: two or three days in office per week, the rest async. Fully remote for mid-level only when the engagement supports it.

EST and PST coverage. We don’t run a 24/7 shop and we don’t page on weekends. Production incidents have an on-call rotation with a written SLA and a real backup. The first time an engineer is on call, a senior pairs with them.

Every engagement runs in the open with the client: public Jira board, weekly demo, written sprint reviews. Engineers see what they ship reach a real operator. The feedback loop is the work.

If we hire you, we’ve read your real code. The interview is a conversation about a system you’ve shipped, what broke in year two, and how you’d build it differently now. If we ask you for a take-home, we pay, and we cap it at three hours.

No “culture fit” panels. The fit is the work. Two of us will pair with you on a real problem from the codebase you’d work on. You ask us the hard questions about how we run; we ask you the hard questions about how you build.

No offshore tier. Every engineer working on a client engagement is hired directly into Sytepoint. The hourly rate reflects that.

Send a note to hello@sytepoint.com with a one-page version of: what you’ve shipped, what your current role is missing, and which practice (AI, platforms, mobile) you’re most drawn to. Link to one piece of production code you’re proud of, public or private. If private, we’ll arrange a walkthrough.

We reply within five business days, including when we’re not a fit. The worst path for both of us is a candidate waiting two weeks for an answer.

Don’t see your shape above? Email anyway. We hire when the shape clicks, not when the req opens. Half our hires have started this way.