Senior Full-Stack AI Engineer — Next.js, Node.js, PostgreSQL, LLM APIs & GEO/ AI Search Platform

We are building a new B2B SaaS product in stealth mode for brands and advertisers. The platform will help companies measure and improve how they appear in AI-generated answers across platforms such as ChatGPT, Claude and Google Gemini. The product will monitor brand and competitor visibility, mentions, recommendations, rankings, sentiment, citations and source quality. It will also diagnose visibility gaps, recommend corrective actions and track whether those actions improve future results. The product name and detailed business information will be shared with shortlisted candidates. We are looking for a product-focused Full-Stack AI Engineer who can help build the initial Release 1 beta and continue with the product roadmap if the collaboration works well. Initial Release 1 scope includes: * Workspace, user authentication and multi-tenant access. * Brand, product, competitor and market setup. * Brand Kit containing products, claims, proof points, audiences and differentiators. * Prompt creation, generation, categorisation and scoring. * Prompt approval and tracking configuration. * AI provider integrations for OpenAI/ChatGPT, Anthropic/Claude and Google Gemini. * Observation-run orchestration across approved prompts. * Background jobs, queues, retries and failure handling. * Storage of raw AI responses and execution metadata. * Extraction of brand mentions, recommendations, rankings, sentiment and citations. * Citation URL and source classification. * Competitor benchmarking. * Share of AI Voice, recommendation rate, mention rate and citation metrics. * Diagnosis of visibility gaps. * Evidence-backed recommendations. * Intervention and outcome tracking. * Dashboard screens, filters and drill-down views. * CSV exports and basic PDF reporting. * Testing, deployment, monitoring and technical documentation. Preferred technology stack: * TypeScript and JavaScript. * Next.js, React and Tailwind CSS. * Node.js backend development. * PostgreSQL, preferably Neon. * Clerk or equivalent authentication. * Redis and BullMQ, or equivalent background-job systems. * Railway or similar cloud deployment. * Cloudflare R2 or equivalent object storage. * OpenAI, Anthropic and Google Gemini APIs. * REST APIs, webhooks and structured JSON outputs. * Zod or equivalent schema validation. * GitHub, GitHub Actions and pull-request workflow. * Docker and environment/secrets management. * Sentry, Langfuse or equivalent monitoring and AI observability tools. You should understand how to build reliable LLM-powered SaaS products, including structured extraction, prompt versioning, schema validation, embeddings, semantic similarity, confidence scoring, evaluation, replay testing, API-cost management, provider fallbacks and malformed-response handling. Experience with GEO, AEO, SEO intelligence, search analytics, recommendation systems, web research tools or citation analysis is strongly preferred. You do not need to be an expert in this field, but you must be able to understand the product logic and convert it into dependable software. The initial architecture should be a maintainable TypeScript-first modular monolith. We do not want unnecessary microservices, Kubernetes, Kafka or excessive infrastructure complexity during the beta. Important engineering expectations: * Treat AI output as data requiring validation, not as automatically reliable truth. * Store raw responses so results can be replayed and audited. * Keep failed, unsupported or unavailable observations separate from zero visibility. * Build provider adapters so individual AI platforms can be enabled or disabled independently. * Include workspace-level authorization and data isolation. * Write tests for scoring, extraction, metrics, permissions and provider failures. * Use synthetic or recorded responses during development wherever possible. * Track API usage, execution status, errors and estimated costs. * Deliver clean, documented and maintainable code. Before development begins, please provide a feature-by-feature estimate based on the scope above. Do not submit only a total number of hours. Please break the estimate into: 1. Project setup and environments. 2. Authentication and multi-tenancy. 3. Brand and competitor management. 4. Brand Kit. 5. Prompt creation, categorisation and scoring. 6. Prompt approval and tracking configuration. 7. AI provider adapters. 8. Observation-run orchestration. 9. Raw response and evidence storage. 10. Mention, recommendation and citation extraction. 11. Competitor benchmarking. 12. Metrics and dashboard. 13. Diagnosis and recommendations. 14. Intervention and outcome tracking. 15. Reporting and exports. 16. Testing and evaluation. 17. Deployment, monitoring and documentation. For every workstream, please provide: * Deliverables. * Frontend hours. * Backend/API hours. * Database and migration hours. * AI/data-processing hours. * Testing hours. * Dependencies. * Assumptions. * Exclusions. * Risks and provider limitations. * Estimated total hours. * Expected completion timeline. The engagement will begin with a paid trial assignment of approximately 15–20 hours. The trial may involve building a small mocked observation pipeline that accepts one brand, three competitors and five prompts, processes sample AI responses, extracts mentions and citations, calculates basic visibility metrics and displays the results in a simple dashboard. Please include the following in your proposal: * Examples of production SaaS products you have built. * Examples of LLM or AI integrations. * Your experience with Next.js, TypeScript and PostgreSQL. * Your experience with queues, background jobs and API integrations. * Your estimated hours for the initial beta. * Your hourly rate. * Your availability per week. * Your preferred development and communication process. * A link to relevant GitHub repositories or working demonstrations. Please do not propose only a chatbot, RAG demo or frontend dashboard. We need an engineer who can own the complete product path from database design and APIs to AI provider integrations, observation workflows, extraction logic, analytics, testing and deployment. This can become a longer-term product development engagement or contract-to-hire opportunity after a successful trial period.