Full-Stack AI Product Engineer for Document-Driven Spatiotemporal Insight Platform

## Job Description We are looking for a strong full-stack engineer, or a small senior team, to build the document-driven AI insight platform. The product combines user profile data, current place/time context, geolocation, symbolic interpretation engines, and AI-generated reports into a guided user experience. This is **not a generic chatbot project**. The core product flow, domain model, architecture, API contracts, data model, AI safety rules, privacy rules, QA process, and release criteria have already been documented in a structured implementation package. The selected developer will receive a canonical implementation pack after contract start and NDA/confidentiality agreement. The work must follow the provided documentation, PR workflow, acceptance criteria, and owner review gates. --- ## What We Are Building The focuses on a guided report flow where a user can generate a personalized “current place and current moment” insight report. At a high level, the platform includes: * User profile and onboarding flow * Consent and notice gates * Location and place snapshot handling * Moment/time snapshot handling * Symbolic interpretation engine layer * AI synthesis/report generation layer * Private report storage and history * Report result UI * Engine breakdown UI * Choice/scenario UI * Privacy-safe logging and audit trail * Admin/reduced-view operational hooks * QA, regression, and release workflow The system must separate deterministic engine outputs from AI-generated explanations. The AI layer must not invent, alter, or override engine-calculated values. --- ## Important Product Principles This project must be implemented as a **document-driven product**, not as an improvised prototype. Key principles: * Build from approved implementation documents. * Every implementation PR must reference the relevant document IDs. * Domain model, API contracts, engine outputs, AI prompts, privacy gates, and release gates must not be changed casually. * Sensitive user data must not be written into logs, analytics, crash reports, queues, or public URLs. * Location permission, platform consent, AI disclosure, and user preferences must be handled as separate concepts. * AI output must be safety-checked and traceable to engine results or approved context. * Feature implementation and feature activation are separate. Some features may exist behind flags but cannot be activated without owner approval. --- ## Ideal Candidate You should be comfortable building a production-oriented from detailed technical documentation. You should have experience with: * Full-stack web application development * Backend architecture and API implementation * PostgreSQL or similar relational databases * Data modeling and migrations * Authentication/session/subject handling * Privacy-sensitive user data flows * Geolocation or location-based product features * LLM / AI API integration * Structured AI outputs, prompt versioning, and safety checks * Frontend guided flows, not just chat UI * GitHub PR workflow, code review, testing, and documentation * CI/CD, staging environments, logging, monitoring, and rollback planning Preferred stack is flexible, but a modern stack such as the following would be suitable: * Frontend: React / Next.js / TypeScript * Backend: Node.js / TypeScript, Python, or similar * Database: PostgreSQL * Queue/Jobs: Redis, BullMQ, Celery, or similar * AI: OpenAI or compatible LLM provider abstraction * Infrastructure: Docker, GitHub Actions, cloud deployment Please propose your preferred stack and explain why. --- ## Scope of Work ### Phase 1 — Project Setup and Architecture Alignment * Review the implementation documentation package. * Set up repository structure. * Set up development, staging, and production environment strategy. * Create initial app shell and backend service layout. * Implement coding conventions, PR template, and test workflow. * Confirm module boundaries and feature gates. ### Phase 2 — Core Domain, Data Model, and Backend Foundation * Implement core domain models. * Implement database schema and migrations. * Implement user/profile/session handling. * Implement consent, notice, and privacy gate logic. * Implement Moment, Place, and snapshot structures. * Implement audit/event/outbox-safe patterns. * Ensure sensitive payloads are excluded from logs and events. ### Phase 3 — Engine Layer * Implement engine input/output contracts. * Implement calendar/timezone/place foundation. * Implement symbolic engine modules according to the provided rules. * Implement normalized signals, compatibility scoring, and versioning. * Implement engine regression tests and fixtures. * Ensure AI cannot modify engine outputs. ### Phase 4 — AI Synthesis and Report Layer * Implement AI context bundle generation. * Implement model routing/provider abstraction. * Implement prompt registry and prompt versioning. * Implement structured report generation. * Implement safety precheck and postcheck. * Implement hallucination controls and evidence references. * Implement private report persistence. ### Phase 5 — API Layer * Implement REST API under a versioned API structure. * Implement standard success/error envelopes. * Implement auth/session subject resolution. * Implement User/Profile/Consent APIs. * Implement Moment/Place/Target/Engine APIs. * Implement Synthesis/Report APIs. * Implement Share/Export/Payment/Admin APIs as gated or inactive where required. * Implement idempotency and API contract tests. ### Phase 6 — Frontend * Implement onboarding and profile flow. * Implement location permission and platform consent flow. * Implement current place/manual place input flow. * Implement Moment input flow. * Implement report generation progress states. * Implement report result screen. * Implement engine breakdown screen. * Implement choice/scenario screen. * Implement report history/private report list. * Implement empty, error, loading, refusal, and disclosure states. * Implement privacy-safe client behavior. ### Phase 7 — Security, Compliance, QA, and Release Readiness * Implement privacy and security guardrails. * Implement consent withdrawal, data deletion, and export flows as required. * Implement audit logging and reduced-view admin access. * Implement unit, integration, contract, E2E, and regression tests. * Implement AI safety and report quality evaluation harness. * Set up CI/CD, staging deployment, monitoring, backup/restore, and rollback procedures. * Prepare readiness evidence package. --- ## Deliverables Expected deliverables include: * Working web application * Backend API and service implementation * Database schema and migration files * Engine modules and regression fixtures * AI synthesis/report generation pipeline * Frontend guided user flow * Admin/reduced-view operational hooks where applicable * Test suite and QA evidence * Deployment configuration * Staging environment * Documentation of implementation decisions * PRs linked to relevant implementation documents * Final handoff package with setup, deployment, and operation instructions --- ## Development Workflow We will use a strict PR-based workflow. Each PR must include: * Related implementation document IDs * Summary of change * Affected domain models * Affected APIs * Privacy/location data impact * AI output impact * Engine calculation impact * Tests performed * Rollback plan * Screenshots or sample outputs where applicable PRs that affect protocol, privacy, engine logic, AI safety, or release gates will require owner review before merge. --- ## What We Are Not Looking For Please do not apply if you intend to: * Build this as a generic chatbot * Ignore the provided implementation documents * Let the LLM invent calculation logic * Store sensitive user data in logs or analytics * Hardcode prompts without versioning * Skip tests or QA evidence * Build only a visual prototype without backend correctness * Make product decisions without owner approval * Activate payment, public sharing, RAG, or admin features without explicit approval --- ## Confidentiality The detailed implementation pack contains proprietary product, architecture, domain, and AI safety documentation. It will be shared only after the contract begins and appropriate confidentiality terms are accepted. Do not upload confidential project documents, prompts, data schemas, or implementation details into public AI tools or third-party services without approval. --- ## Application Instructions Please include the following in your proposal: 1. A brief summary of your relevant experience. 2. Links to similar full-stack, AI, location-based, or privacy-sensitive products you have built. 3. Your proposed technical stack. 4. How you would structure the implementation milestones. 5. How you approach document-driven development. 6. How you handle LLM safety, structured outputs, and hallucination prevention. 7. How you handle sensitive user data, location data, logging, and audit trails. 8. Your recommended team structure if more than one person is needed. 9. Your availability and expected weekly hours. 10. Any assumptions or risks you see from this job description. Please start your proposal with: **“I understand this is a document-driven AI product, not a generic chatbot.”** This helps confirm that you read the posting carefully. --- ## Suggested Contract Type Milestone-based fixed price or hourly with weekly deliverables. Please propose your preferred structure. Suggested milestone structure: 1. Documentation review, architecture alignment, and repo setup 2. Domain model, database schema, and backend foundation 3. Engine implementation and regression tests 4. AI/report generation pipeline 5. API completion and contract tests 6. Frontend implementation 7. Security/compliance hardening 8. QA, staging deployment, and release readiness package