Senior Full Stack Engineer

*** EQUITY FIRST, REVENUE & COMPENSATION TO BEGIN WITHIN 2 WEEKS OF LAUNCH*** About Arcana Arcana is a decision-intelligence platform that stress-tests high-stakes decisions, surfaces blind spots, and produces auditable, source-tagged outputs through a structured six-stage protocol. We build for two markets underserved by enterprise tools: growing SMBs, and research, policy, and think-tank organizations that need rigour and an audit trail they can defend. A working prototype exists and is in private concierge-led beta. What it does not yet have is a production spine: the system needs to move from a promising prototype to reliable, multi-user software. That is the job. WHY THIS ROLE EXISTS The role We are looking for a hands-on senior engineer to take Arcana from prototype to a production-ready MVP — and to build it properly the second time. This is a builder role, not a manager role: your hands are on the keyboard. You will own the codebase end to end, close the reliability gaps that currently block real users, and ship something we can put in front of paying beta customers. You will work directly with the founder and the Chief AI & Strategy Officer, who owns AI architecture direction. You own how it gets built and whether it holds up. As the person who ships the first real version, you have unusual influence over the architecture, the stack decisions, and how the engineering function grows from here. THE CONCRETE WORK Your first 90 days The prototype was built to prove the idea, not to survive real load. Six issues currently block self-serve access, and clearing them is your opening mandate: Async that doesn't block — migrate synchronous HTTP calls to a proper async client (requests → httpx) so the pipeline stops blocking.Durable storage — no persistence today; move working state into a real database so nothing is lost on restart.Rate limiting — protect the system and control cost per decision-run.Error recovery — checkpointing so a long decision-run can resume instead of starting over.Data validation — validate inputs and outputs so bad data can't quietly corrupt a run.Caching — add a caching layer to cut latency and repeated model spend.From there: stand up multi-tenant, multi-user architecture with authentication and role-based access — the gate to self-serve — and get a demo-ready, testable MVP live for our beta customers. THE STACK What you'll build with Backend — FastAPI / Python (async), with a RAG-based agentic architecture on the Anthropic Claude API.Frontend — React.Data — PostgreSQL (migrating from SQLite) with Redis for caching.Infra — cloud deployment, CI/CD, staging and production environments — enough to ship and keep it running.Integrations — planned read-only connectors (e.g. web-scrape/enrichment, market and economic data sources) feeding the decision pipeline. REQUIREMENTS What you must bring Full-stack depth — Senior full-stack experience — you can own both the Python backend and a React frontend without hand-holding.Python / FastAPI / async — Real, hands-on production experience with Python and FastAPI, and genuine command of async (this is the difference between the prototype and a product).AI / LLM engineering — Direct experience building with LLMs — RAG pipelines, agentic/tool-use patterns, retrieval, and prompt/pipeline design. Anthropic API experience is a strong plus. This is the hardest and most important skill for this role.Databases — PostgreSQL and schema/migration work; Redis or comparable caching.Testing discipline — You write automated tests by habit. Our current gaps exist precisely because the prototype was never tested; we need someone who won't repeat that.Ships to production — Enough deployment, CI/CD, and cloud know-how to take a feature from your machine to production. Not a specialist — just self-sufficient. BONUS Nice to have Startup 0→1 — Prior 0-to-1 / founding-engineer experience at an early-stage startup — comfort with ambiguity and moving fast without breaking trust.Self-hosted inference — Experience with self-hosted / open-source inference (e.g. vLLM, open-weight models) — a post-MVP cost-optimization item on our roadmap.Security — Security-minded architecture and experience handling sensitive client data responsibly.Leadership trajectory — Interest in growing into a technical leadership role as the team and funding grow. An honest note on stage and compensation Arcana is pre-revenue and pre-funding. Early compensation is weighted toward meaningful founding equity, with cash compensation that begins on our first funding close or a defined revenue milestone. If you need a full market salary from day one, this stage isn't the right fit yet — and we'd rather say that plainly up front. What we can offer instead: real ownership, a clean codebase you get to shape from the ground up, concrete and meaty problems, and a direct line to the founder. Equity is tied to delivery milestones and vests over time. HOW WE WORK Working terms Remote — Remote-first, Canada-based. Ontario / GTA is a plus for the occasional in-person session but not required.IP & confidentiality — A signed NDA and full IP assignment are required before any access to the codebase. This is non-negotiable and applies to everyone who touches the code.Jurisdiction — Ontario, Canada. We work with Ontario counsel on all founding-team agreements. NEXT STEP How to apply Send a short note on what you've built (links to shipped work, GitHub, or a relevant repo are worth more than a long CV), your experience with LLM/RAG systems specifically, and what draws you to an early-stage build. We move quickly for the right person. Contact: Alexandra (Lexi) Kubrak, Founder & CEO — lexi@enigmathinkers.com