Principal Software Engineer, AI Systems

About Promobile

Promobile is an experiential marketing agency that designs and produces large-scale brand activations for major consumer brands. To scale what we do, we're building an AI-powered software platform that plans, prices, and operates those activations — used first by our own team, then offered to the wider experiential marketing industry.

You'd join the small software team behind that build, backed by an operating agency with real customers, real budgets, and deep domain expertise. We ship production AI systems, not demos.

What We're Looking For

You've shipped at least one production system that calls an LLM and reasons about the output structurally. Not chat completions or one-shot prompts, but a real pipeline with prompt caching, Zod-validated structured outputs, evaluation harnesses, and the ability to debug why a model returned the wrong shape three weeks after you shipped it.

On top of that, you've shipped full-stack TypeScript products in production. You've debugged a broken RLS policy. You know when to reach for a database function vs. application logic, when something should be configurable vs. static state. You have opinions about data fetching patterns and can explain why optimistic updates matter for UX.

You're comfortable being the most senior technical voice in the room. You'll push back on bad ideas, but you'll also execute on decisions even if it means adding complexity on the backend so the user gets the best experience.

Experience: 6+ years building production web applications, with at least 1 year of that on production AI/LLM systems.

Not The Right Fit If

• You've used LLM APIs but haven't designed a production pipeline around them
• You're transitioning from frontend-only or backend-only work and want to "grow into" full-stack
• You need an engineering manager, a senior peer to learn from, or structured onboarding to ramp up
• This would be your first Senior, Lead, or Principal role

Responsibilities

• Own delivery of a modern SaaS web application with AI integrations
• Architect modern web features deployed to serverless infrastructure and cloud (edge) functions for long-running tasks. Prioritize correctness, performance, and flexibility
• Model complex problems in a relational database, manage schema migrations, optimize query performance, and secure data with RLS policies
• Write production-quality TypeScript across the full stack: Next.js App Router (React) with TailwindCSS on the front end, Supabase Edge Functions (Deno) on the back end, and PostgreSQL with RLS for data integrity
• Build and maintain the AI pipeline, an extensible orchestration system (Claude Haiku + Sonnet) that generates structured outputs with prompt caching and Zod-validated schemas
• Implement and iterate on UI features using our Glow UI design system, making sure every interaction feels fast and intentional
• Manage CI/CD pipeline via Github Actions and Vercel webhooks
• Participate in and challenge product and UX discussions. We operate with full transparency, so you're expected to have opinions on what to build and why
• Develop test suites that provide important feedback in a fast-moving startup environment. It's important to move fast, but also move forward, and a strong test suite prevents regressions

Culture

We're a small team that punches far above our weight. A few things define how we work:

We have strong architectural opinions. Enforcing strong architecture allows us to move fast without reinventing the wheel on every feature. Pick a direction and stick with it until it needs upgrading.

We move fast without cutting corners. Fast means clean builds, clean lint, and working tests, not skipping them. Surprises cost time, and minimizing them upfront allows us to move fast.

We don't hide from bad news. If something is broken, architecturally wrong, or likely to cause problems at scale, we address it head on and stay open to change. Prioritize the user always.

We treat AI as a collaborator. We use Claude heavily in our development workflow (Claude Code, Cowork, and in-product pipelines). You should be comfortable prompting, evaluating, and iterating on AI-generated code rather than rubber-stamping it.

Technology

• Next.js 16 (App Router)
• React 19 + Tailwind CSS v4
• TypeScript
• Supabase (PostgreSQL)
• Supabase Auth (JWT)
• Deno (Supabase Functions)
• Anthropic Claude API (Haiku + Sonnet)
• SWR + SWR Mutation
• Vitest

Logistics

• T4 (Canada)
• Montreal

Pay: $140,000.00-$180,000.00 per year

Application question(s):

• Describe the most complex production system you've designed and shipped end-to-end. What was the hardest architectural decision you made, and what would you do differently now?
• A link to something you've built (a GitHub repo, deployed product, technical writeup, or portfolio piece). Add one sentence on what we should look at and why.

Work Location: In person