Description We’re transitioning an existing SaaS product (Traqplan) into CHELT AI — an AI-assisted governance and decision-support platform built around structured “registers” (actions, risks, decisions, issues, etc.). This is not a greenfield project and not a simple chatbot wrapper. We need a senior engineer (or small team) who can work inside an existing codebase and deliver an initial production-quality foundation, then continue iterating. Current Stack (Existing System) Angular (web) Node.js API PostgreSQL AG Grid (primary data grids) Mobile app via Capacitor wrapper Key Product Principles (Non-Negotiable) The database contains structured records in registers (actions/risk/decision registers etc.). AI can interpret user intent and generate proposals, butmust not directly access or write to the database. Updates must be applied through deterministic backend services/APIs with: explicit user confirmation/approval where needed auditability and traceability no silent mutations and no “hidden state” stores for AI convenience Initial Scope of Work (What You’ll Build First)1) Product transition: Traqplan → CHELT Implement the rebrand and restructure UX flows for the CHELT model Add dark palette as default and replace Traqplan branding with CHELT AI Introduce the new UX structure: Chelt Hub = main app entry point (rebranded existing “virtual c-suite”) Chelt Governance Vault = dedicated chat/AI interface that opens in aseparate browser window Add admin ability to manage/switch personas (used for AI interaction and experience)2) Block #1 (highest priority): Natural language CRUD on register records Users can use plain English to create/update/query register items (actions/risks/decisions/etc.) Implement the safe pattern: interpret intent → generate structured change proposal → user confirms → backend applies via API → log/audit Ensure register schemas are treated as contracts; changes are versioned/history-preserving3) Block #2: Analysis / Insights from register data Enable AI to generate insights grounded in register data (cross-register Q&A) Responses must be explainable/traceable to underlying records (no unsupported hallucinations)4) Block #3 (next): Agents that “do things” (governed automation) Build the foundation for an “action module” where agents can execute defined tasks/workflows Strong governance requirements: approvals, run logs, identity (“run as”), failure handling/retries Bonus if you’ve worked with workflow engines (e.g., n8n) in a governed/enterprise context Infrastructure / Deployment We will support multiple environments and domains: UAT:uat.chelt.ai Multi-tenant production:app.chelt.ai Single-tenant enterprise:[ORGNAME].chelt.ai Cloud is expected to be AWS-first (open to final confirmation) Who We Want Senior backend/architecture strength (Node/Postgres), comfortable in Angular systems too Experience shipping AI features into real products with governance and safety constraints Comfortable refactoring an existing codebase (not rebuilding everything) Opinionated about clean boundaries, auditability, versioning, and permissions Interested in continuing beyond the first phase (long-term product work) To Apply Please include: 1–2 examples of production systems you built/refactored (especially with AI or automation) How you would implement “intent → structured proposal → confirm → apply → audit” safely Any relevant experience with multi-tenant SaaS and/or workflow orchestration