Full Stack AI Engineer

Full-Stack AI Engineer — React + LLM Integrations Governed AI platform · Australia Type: Contract, ongoing (20–40 hrs/week to start, scope to grow) Location: Remote. Must overlap at least 4 hours with AEST business hours. Start: Immediately About BIQc BIQc (biqc.ai) is an Australian governed-AI platform for SMBs and enterprises. We give owners and teams a ChatGPT-grade chat (Ask BIQc) that is connected to their real business systems — with scoped tool access the customer approves, per-user permissions, instant revocation, and a full record of every AI interaction. We sit on top of foundation models (Anthropic, OpenAI and others). We do not compete with them. Staff use AI through BIQc; BIQc controls policy, permissions, auditability and business context. The product is live with paying customers. We’re hiring because the roadmap is moving faster than a small senior team can ship. What you’ll work on first This is not a greenfield role. You’ll join a live product and make the next customer-facing slices real, visible and stable. Ask BIQc, at ChatGPT/Claude quality. Composer, tools, plugins, connections and history all have to be findable. If a customer can’t find a feature, it doesn’t exist. Plugins and actions. Let Ask BIQc take approved actions in the user’s apps — connect, confirm, run, revoke. You’ll extend and harden the tool-calling path, not invent a parallel one. Company connections. Owner/admin connects a system once for the workspace. You’ll work on connect/reconnect, sync health, and the “what BIQc can see / what this person can do” surfaces (My connections, Role permissions, Data rules). Native integrations. Email, calendar and a small set of Australian and SaaS tools we support directly (including field-service and Zapier). OAuth, webhooks, retries, sync state — production, not demos. Stabilisation. The React app grew fast. Part of this role is untangling shared chat/settings/connection UI properly so the next feature doesn’t add another knot. Later work includes turning research modes (finance, people) into first-class surfaces, and scheduled briefs. Day one is the chat, the plugins, and the connections. You 4+ years shipping production web apps. Strong React on the front end; solid backend/API work. Hands-on with LLM APIs in production: tool/function calling, streaming, structured outputs, prompt/context management, cost control. You’ve integrated third-party SaaS APIs and webhooks (OAuth, rate limits, retries, sync state). You’ve built or maintained a chat UI with tools, history and multi-user state. You can work from a Jira ticket, record a short Loom walkthrough per delivered fix, and check your own work against ChatGPT/Claude before calling it done. Clear written English. You’ll talk directly with the founder and the design lead. TypeScript is a plus. So is having shipped this kind of product before, not only a weekend wrapper. Strong plus MCP (Model Context Protocol) servers or clients Unified APIs or agent/tool platforms (e.g. Merge.dev) HubSpot, Zapier, field-service or accounting APIs Auth, RBAC and audit logging for multi-tenant SaaS Australian or APAC time zone How we work Small senior team: product, engineering, design. Daily engineering sync. We manage outcomes, not hours. Jira, Slack, Loom, Figma. Quality bar: BIQc must look, feel and work like ChatGPT and Claude. Customer-facing copy has to pass the “would your sister understand it” test. No vendor jargon in the product. To apply Reply with: One production app where you integrated an LLM with tool calling — what it did, what broke, what you’d do differently One example of untangling an unstable front end without a rewrite Your weekly availability and AEST overlap Your hourly rate Short and specific beats long and generic. Applications that open with “Dear Hiring Manager” and a template will be skipped.