Software Engineer, Platform

Siena is the Agent of Record for CX, built for consumer brands like Spanx, FIGS, Brooklinen, and Grüns.

It runs as a connected multi-agent stack on top of a shared intelligence layer: agents that sell, support, listen, handle social and voice across the entire customer journey, turning every conversation into signal you can act on.

We are a small, fully distributed, global team. We have built leverage into how we work, which is what lets a team this size move at the speed we do.

We hire athletes. Everyone here owns real outcomes, defaults to writing over meetings, communicates with directness, and acts the moment they see something worth acting on. Low ego, high standards, fast clock.

If you want to set the bar rather than follow one, keep reading.

You take the hardest problems customers hit in production and solve them end to end: a webhook that stopped firing, an agent filling the wrong parameter, a report that won't reconcile. You find the real cause, work the customer to resolution, then build the automation so it doesn't come back. This is the front door to deployment and platform work at Siena, with a clear path into Forward Deployed Engineering or deeper platform work.

• Own hard problems end to end: trace, reproduce, fix, or hand engineering a clean reproduction. Get on the call when that's the fastest path

• Debug the agents and the prompts. Diagnose whether a misfire is prompt, knowledge base, config, or model, then fix it so it holds under real traffic

• Build so it doesn't happen twice. For every fix, ship the runbook or tool that stops it recurring

• Close the loop to Product. Surface patterns, not one-offs, backed by metrics

• Raise the team's floor. Lift what the whole team can resolve, not just what you can

• Know where your scope ends. Escalate genuine product bugs and outages with a clean reproduction, not a shrug

• Within 90 days, name the top recurring failure patterns and attack them at the root, not just patch symptoms

The Bar: If you're missing any of these, this isn't the role for you.

• You solve problems end to end. Given a customer, some logs, and a vague symptom, you find the break and can tell a customer-side cause from a product bug

• You're fluent with AI agents and prompts. You've built or debugged LLM workflows and know how they break: prompts, knowledge bases, retrieval, the gap between test and live

• You're excited to build automation, not just fix what's in front of you. You think in leverage. This is the signal that matters most

Strong Signal: Not deal breakers, but what separates a good hire from a great one.

• You like talking to customers. You'll get on a call and resolve the real problem live, not just the reported one

• Comfortable in our core stack (Node.js, Type Script, React, Express) or close enough to ramp fast

• Fluent with AI coding tools (Claude Code or equivalent) to script, clean data, or navigate a new codebase fast

• You own problems nobody assigned you, and you make the people around you better through docs and walkthroughs

Nice to Have: Bonus, not required.

• Observability tooling (Grafana, Honeycomb, Sentry)

• Internal tooling that turned manual work into a system

• Ecommerce or DTC exposure

Not a Fit If:

• You stop at "it's not our bug" instead of finding the real cause

• You'd rather build the fix than talk to the customer who hit it

• You want a backlog of well-scoped tickets, not ambiguous production fires

Real ownership. You will own problems, not tasks, with the autonomy to solve them your way and the trust to do it without a manager in your inbox.

Work that ships and matters. What you build reaches real consumer brands fast.

A category we are creating. The Agent of Record is new ground, with unsolved problems and cutting-edge models, not a crowded field to compete in.

A team that respects your judgment. Globally distributed, async-first, low bureaucracy. We hire adults and treat you like one.

Great salary plus equity or stock grants. Own a piece of what you build.

Learning budget. If you are growing, so are we.

AI-fluency by default. Few places will push your work with production AI further or faster.

The people who thrive here are curious, customer-obsessed, and take ownership without being asked. They fix problems first and explain later. They're direct about feedback - both giving and receiving it - because they care more about getting things right than being polite.

They maintain high standards while moving at startup speed, and they build real…