Full-Stack Developer, AI Applications

Full-Stack Developer, AI Applications Savvier | Product Engineering | Toronto or remote in Canada Company Savvier, besavvier.com Location Toronto, Ontario. Remote-friendly within Canadian time zones, with occasional in-person sessions. Reports to Engineering Lead, working directly with the CEO and Chief Product Officer About Savvier Savvier puts senior-level marketing expertise into the workflow itself. Most AI tools are only as good as whoever is prompting them. We built something different: 50+ workflows created by proven strategists and creative directors, backed by persistent brand context, quality controls, and full auditability. Marketing teams direct the work. Savvier provides the strategic framework. Teams at HelloFresh, Deloitte, Transit Technologies, and Edward Jones use the platform today. Our enterprise business is growing across consulting firms, financial services, and enterprise brands. We are a small, senior team based in Toronto. About the role We are looking for a Full-Stack Developer to own and extend the product surfaces where our customers interact with the AI, and to build what comes next. That includes the workflow experience, the collaborative editor, the brand context system, and the APIs that connect them. You will work across both sides of the stack: building interactive interfaces for AI-driven output on the front end, and designing the orchestration logic, data models, and service integrations that power them on the back end. You do not need to have trained a model. You do need to be comfortable building software where the output is probabilistic and the failure modes are different from traditional software. We sell into enterprise, so multi-tenancy, permissions, audit trails, and security are core to the product and actively evolving. You will have a direct hand in how they grow. What you will do • Build features end to end, from the React front end through the API layer to the database, plus the Python services behind our agents. Streaming, long-running jobs, and multi-agent experiences are the norm here. • Design interfaces for AI output that set the right expectations: token-by-token streaming, visible progress through multi-step workflows, intermediate states, source citations, and clear paths to review, edit, and approve. Much of this lives in our collaborative document editor. • Extend the orchestration layer behind our workflows: prompt and context assembly, brand context retrieval, tool calling, model routing, retries, and graceful fallback when a provider degrades. • Extend our brand context system so every project starts with the right positioning, voice, audience, and prior work loaded automatically. That covers the data model, the document pipeline that feeds it, and the hybrid retrieval layer on top. • Design versioned APIs and data models that support multi-tenancy, role-based permissions, and auditability from the start. Every request in our system resolves to an organization. • Extend the enterprise controls our buyers require: single sign-on, granular access control, audit logging, and the controls behind our SOC 2 work. • Instrument what we build: tracing, evaluation harnesses, cost and latency monitoring, and the checks that tell us whether a change made the product better or just different. • Maintain quality across the stack: automated tests across the TypeScript and Python sides, code review, CI/CD, and postmortems when something breaks. • Work directly with product and our workflow authors to turn expert marketing methodology into software that holds up under real use. What you bring If you meet most of the following, we want to hear from you. • Four or more years building production web applications, with real ownership across both front-end and back-end. • Strong TypeScript and modern React, including the Next.js App Router, server components, streaming responses, and state management in interactive applications. • Strong back-end skills in TypeScript and Python. Depth in one and real working comfort in the other, plus REST API design, relational data modelling in PostgreSQL, and queue-backed asynchronous processing. • Hands-on experience integrating LLM APIs into a production product: structured output, tool calling, streaming, context window management, and cost control. • Working knowledge of retrieval patterns: embeddings, chunking strategy, vector and hybrid search, and clear judgement about when retrieval is and is not the right answer. • Comfort with cloud infrastructure and modern delivery practices: containers, CI/CD, and infrastructure as code. We run on Google Cloud and Vercel; equivalent depth on AWS or Azure transfers cleanly. • Experience building multi-tenant SaaS: org-scoped data models, role-based permissions, and auditability as design constraints rather than later additions. • A practical approach to quality in systems where the same input does not always produce the same output. • Clear written communication, and the self-direction to work well on a small distributed team. • Familiarity with AI-assisted development tools and a willingness to integrate them into your daily workflow. • A degree in computer science or a related field is welcome but not required. Demonstrated ability through work, open source, or professional experience carries equal weight. Bonus points • Experience with agent frameworks and orchestration tooling such as LangGraph, LangChain, or the Vercel AI SDK. • Experience with LLM observability and evaluation tooling, whether vendor or custom built. • Background in AI safety and reliability: guardrails, prompt injection defence, PII handling, or human-in-the-loop review design. • Experience with enterprise buyers, including SOC 2, vendor security reviews, and penetration test remediation. • Marketing, agency, or martech background, or genuine curiosity about how marketing teams work. • Design sensibility and a track record of building interfaces people find pleasant to use. Our stack TypeScript, React, Next.js, Node.js, Python, PostgreSQL, Neo4j, Google Cloud, Vercel, Docker, GitHub Actions. Our workflows are built on Anthropic Claude with a multi-agent architecture. We expect you to help us evolve this list. What we offer • A small team with a short path from idea to production. • AI tools are part of how we work. You will have access to the ones you need. • Work that goes live. Nothing you build sits in a backlog waiting for a quarterly planning cycle. How we hire We aim to move from application to offer in two to three weeks. 1. Introductory call with the hiring manager, 30 minutes. 2. Technical conversation about your past work and how you approach building with AI, 60 minutes. 3. Practical exercise: a paid, time-boxed take-home or a collaborative working session, your choice. 4. Conversation with the founders, 45 minutes. 5. Reference checks and offer. Equal opportunity Savvier is an equal opportunity employer. We welcome applicants of every background, and we make hiring decisions based on the ability to do the job. If this role interests you and you meet most of it, please apply. To apply, send your resume to hr@besavvier.com