AI Engineer (AI Responses will be ignored)

AI Engineer: Multi-Agent AI Microservice for SaaS CRM (All AI responses are ignored) Project Overview We are developing a CRM and need to integrate an AI intelligence layer for an existing multi-tenant SaaS CRM. We are seeking an experienced AI and backend engineer to develop a reusable agentic AI microservice and implement eight specialized AI agents. This is primarily a backend and AI architecture engagement. You will not be responsible for frontend or UI development and will collaborate with a full stack dev to make sure your work integrates into the product. The project is expected to be completed within six weeks We are looking for an efficient and highly capable developer who can work within an existing codebase, make pragmatic architectural decisions, and deliver a functional, maintainable, and extensible system. What You Will Build The primary deliverable is a centralized AI orchestration service that allows the CRM to: • Securely retrieve authorized CRM records. • Understand user intent. • Select and invoke the appropriate AI agent or tool. • Use structured and unstructured CRM context. • Apply agent-specific system prompts and workflows. • Return validated, structured outputs. • Propose CRM actions for user review and confirmation. • Support continued conversations through the CRM’s AI chat. • Run scheduled and event-driven agents. • Allow additional agents to be added without rebuilding the core service. The LLM must not have direct or unrestricted database access. Authentication, tenant isolation, record permissions, tool execution, validation, and CRM write actions must remain controlled by the backend. Initial AI Agents You will implement the foundation and workflows for: 1. Daily Briefing Agent Produces prioritized daily and record-specific insights based on tasks, deadlines, communications, opportunities, commissions, and CRM activity. 2. Email Creator and Polisher Generates or improves emails using user instructions and authorized CRM context. 3. CRM Record Generation Agent Generates proposed tasks, projects, project task plans, opportunities, commissions, and other supported CRM records for user approval. 4. Athlete Market-Value Change Alert Agent Explains meaningful changes in proprietary athlete market valuations and recommends appropriate next actions. 5. PR and Media Sentiment Agent Uses web and news search and the Grok API for X/Twitter monitoring, sentiment analysis, rumor detection, urgency classification, and recommended actions. 6. Concierge Form Agent Converts unstructured user requests into validated form JSON, identifies missing fields, and asks targeted follow-up questions. 7. Image and Social Content Agent Produces sports-business image-generation workflows, captions, platform sizing, branding options, and iterative image-edit instructions. 8. Legal Document Drafting Agent Drafts representation agreements, limited mandates and authorizations, offers, and counteroffers using approved templates, clause libraries, CRM context, and open-ended LLM drafting. The workflow must support missing-information questions, revisions, comparisons, versioning, and editable document export. Core Technical Responsibilities • Design the multi-agent orchestration architecture. • Build an extensible agent registry and intent-routing system. • Develop reusable CRM retrieval tools. • Implement tenant and record-level permission enforcement. • Build structured-query and semantic/vector retrieval workflows. • Create versioned system prompts and agent prompts. • Engineer structured prompt workflows for each agent. • Define and validate JSON output schemas. • Implement response repair, retries, and controlled error handling. • Create proposed-action preview and confirmation workflows. • Implement scheduled and event-driven execution. • Integrate approved LLM, search, image-generation, and external API providers. • Add AI run logging, prompt versions, tool-use records, errors, and relevant performance metadata. • Create repeatable evaluation cases and automated tests. • Document APIs, prompts, agent registration, configuration, deployment, and known limitations. • Commit all source code and work product to the Company’s private GitHub repository. Required Experience Applicants should have strong experience with: • Production LLM applications. • Agentic AI or multi-agent orchestration. • OpenAI APIs and tool or function calling. • Retrieval-augmented generation. • Prompt and system-instruction engineering. • Structured LLM outputs and schema validation. • Vector search and embeddings. • Secure multi-tenant SaaS architecture. • Node.js, TypeScript, and Python backend development. • MongoDB and preferably MongoDB Atlas Vector Search. • REST APIs or equivalent service architecture. • Background jobs, scheduled processes, retries, and logging. • GitHub-based development and technical documentation. Experience with AI-generated documents, image-generation APIs, CRM applications, or legal drafting systems is strongly preferred. How to Apply Please provide: • A brief explanation of your experience building agentic AI or tool-calling AI systems. • Links or descriptions of relevant projects. • Your experience with RAG, vector search, structured outputs, and multi-tenant permissions. • Your proposed architecture for routing CRM requests among specialized agents. • Confirmation that you can commit to the six-week timeline and fixed budget. • Confirmation that you are comfortable assigning all project intellectual property to the Company.