Senior Product Engineer — Platform Hardening & Build-Out

Senior Product Engineer — Platform Hardening and Build Out Company: Performance marketing agency (Australia) Location: Australia (remote friendly; some overlap with AEST hours) Type: Three month contract, with optionality for extension — front loaded with a review and harden mandate, then ongoing build out Comp: [range TBD] Why this role exists We have built an internal platform that is already substantially live. We now want one senior engineer, embedded in the live product, to do two things, in order: 1. Pressure test what exists. Run a deep, independent architectural review, surface the real risks, and harden the platform where it needs it. 2. Finish it. Own and build out the remaining surface area of the product to a reliable, scalable v1+. This is a high ownership role. You will not be handed a backlog of tickets — you will be handed a real, working, non trivial system and asked to make it bulletproof and then take it the rest of the way. We are looking for someone who can hold the whole thing in their head, reason about it commercially, and move fast without breaking the integrity of a stateful production system. What the platform is We are a performance marketing agency. Delivering one campaign takes roughly one hundred forty to one hundred eighty hours, about two thirds of it coordination overhead — handoffs, status chasing, QA triggers, approvals, and reporting. That workflow previously ran as a two hundred thirty two step manual checklist in a project management tool that documented the work but executed none of it. The platform replaces the checklist with software: a stateful, multi stage state machine that executes most of those steps automatically and stops at a defined set of human decision gates. Work is routed across automation tiers — deterministic operations, visual/QA automation, LLM assisted drafting (human reviewed before it ships), and the manual tasks that remain. What it includes today: A relational Postgres data model with row level security A tiered background job system with concurrency and retry semantics Operator cockpit UI Automated visual QA A communications layer and client onboarding Integrations with advertising platforms, messaging providers, and workforce systems The work it automates is skilled human coordination time — the agency's most expensive and least scalable input. Removing it lowers cost per campaign and frees senior staff for billable work. That is how we prioritise. Current state: much of the platform is built and running. This role hardens what exists and ships what doesn't — not greenfield, not maintenance. The role, in two phases Phase 1 — Architectural review and hardening (first priority) Conduct a deep, independent architectural review across the full stack: data model and migrations, access control, the state machine, the background job/concurrency layer, the integration surfaces, the API layer, the front end, and the operational/deployment posture. Surface and rank the real risks — correctness and data integrity issues, security and row level security gaps, concurrency/idempotency hazards, race conditions and state consistency bugs, performance cliffs, operational fragility, and the tech debt that will actually bite. Prioritise findings by severity and commercial impact. Execute the hardening: refactor safely at scale, close the gaps, add the tests and guardrails, and leave the platform measurably more robust, secure, observable, and maintainable than you found it — without taking the system offline or rewriting for its own sake. Phase 2 — Platform build out Own and ship the remaining platform surface area to a dependable v1+: workflow automation, external integrations, LLM assisted features (with the right human gates), reporting and analytics, and operator facing tooling. Make sound architectural calls as you go. Become the engineer who can both harden and build — the technical owner of the platform's reliability and its roadmap delivery. What we need from you Core engineering (table stakes for a senior product / DevOps leaning engineer) Senior full stack / product engineering, TypeScript end to end. Strong with a modern React framework (App Router, route handlers, server components) and production front end work. Deep PostgreSQL: schema design, safe migrations, row level security, query performance and indexing, and data integrity thinking. (We use Postgres; database management experience is a plus.) Backend / distributed systems fundamentals: background job and queue systems, concurrency control, idempotency, retries, and eventual consistency, and reasoning correctly about state machines and failure modes. API and integration engineering: building reliable integrations against third party APIs and webhooks — handling auth, rate limits, partial failure, and reconciliation. DevOps / platform: CI/CD, serverless deployment, environments, secrets management, observability, and production reliability practices. Security: authentication/authorisation, RLS, data protection, and a secure by default instinct — you reach for well tested libraries over bespoke crypto/sanitisation. Quality discipline: automated testing, strong typing, and verification before done — you prove behaviour, you don't assert it. Architectural judgement: you can walk into a substantial unfamiliar codebase, build an accurate mental model quickly, threat model it, and refactor it safely. This is the single most important core skill for Phase 1. The two things that genuinely set this role apart These are not nice to haves. They are why this role is different from a standard senior engineer posting, and we will weight them heavily. 1. Commercial acumen You think like an owner of the business, not just the codebase. You connect engineering decisions to margin, billable hour savings, time to value, and client outcomes — and you can articulate that connection. You prioritise ruthlessly by ROI. You know what good enough to ship means commercially, and you don't gold plate things the business doesn't need. You make pragmatic build vs buy decisions and can defend them. You can talk to non technical operators, account managers, and leadership, understand the agency's economics, and translate fuzzy commercial goals into the right technical bets. 2. AI native engineering sophistication You are genuinely, demonstrably sophisticated with AI coding tools — this is core to how we work, not a novelty. Fluent with modern agentic coding tools and workflows (e.g. Claude Code, Cursor, subagent orchestration, spec- and plan-driven development), and you use them to compress the timeline on architectural review, large-scale refactors, and build-out. You understand AI tooling's failure modes and pair generation with rigorous verification and evals — you move fast because you have a discipline around correctness, not at the expense of it. Strong context and prompt engineering: you know how to feed these tools the right information and structure work so they produce trustworthy output on a real, stateful codebase. You can also build AI into the product responsibly — LLM assisted features with appropriate human in the loop gates, because shipping unreviewed model output to clients is not an option. Bluntly: we want someone operating at the frontier of AI assisted delivery, who can make a small team perform like a much larger one. Nice to have Domain exposure to agency / martech / adtech, campaign operations, or marketing analytics. Experience with tiered job runners, browser automation/visual QA, vector search work, or data visualisation. Comfort working in a high context, documentation heavy, quality gated engineering environment. What success looks like First ninety days (Phase 1): a credible, prioritised architectural risk assessment that leadership trusts, plus the highest impact risks actually remediated. The platform is measurably more robust, secure, and maintainable — and we know, with evidence, where the remaining sharp edges are. Beyond (Phase 2): consistent, high quality delivery of the remaining platform, with commercial judgement visible in what you choose to build and in what order. You become the trusted technical owner of both the platform's integrity and its forward roadmap. How to apply Please apply directly by submitting an Upwork proposal. In your application message, please include a real example of you using advanced engineering tooling to do serious engineering work (a refactor, an architectural review, a non trivial feature), and one decision where you traded engineering perfection for the commercially right outcome. These two stories tell us more than a standard text resume does. Note: To ensure compliance with platform terms, all messaging, screening, and contract administration will be handled entirely within Upwork.