AI Engineer

MGI Inc. is building an internal AI automation platform that powers production AI agents used across project management, accounting, contracting, and IT. This hybrid role in Meridian, ID focuses on designing and maintaining AI agents that complete real work end to end, from API integration through deployed agents employees use daily.

What You’ll Work With

• A multi-agent system with several named AI agents, each with its own role, tool set, and company identity serving different departments.
• Chat-based agents employees message directly in Microsoft Teams to receive substantive answers.
• Retrieval system with millions of indexed documents, including company email and attachments, construction codes and standards, regulations, and project documentation.
• Deep third-party integrations including Procore, Microsoft 365 and Graph API, SharePoint, payroll and accounting systems, contracting data sources, and security tooling.
• Scheduled automation jobs (well over a hundred) for syncs, monitoring, reporting, and alerting.
• Local compute fleet for vector search, model routing, and local inference alongside frontier API models.
• Platform layer: OpenClaw, with agent behavior defined through configuration and markdown, while Python implements the tools each agent can call.
• Roughly 200,000 lines of Python powering agent tools and integrations.

Responsibilities

• Design, develop, and refine AI models using frameworks such as TensorFlow and other machine learning tools to solve complex problems.
• Implement NLP techniques for data extraction and analysis from unstructured sources.
• Use big data systems like Hadoop and Spark to process large datasets for predictive modeling analysis.
• Partner with cross-functional teams to integrate AI models into cloud-based platforms using AWS and machine learning cloud services.
• Perform statistical analysis, model training, evaluation, and validation to ensure accuracy and robustness.
• Build scalable data pipelines with ETL processes, Talend, and SQL databases to support ongoing AI initiatives.
• Deploy AI models into production environments with a focus on evaluation, monitoring, and continuous improvement.

Requirements

• OpenClaw or a comparable agent framework.
• Hands-on experience building agents is required.
• Comfort with agents and session management, including skills system, tool definitions and tool permission policy, gateway configuration, scheduled agent tasks, agent memory and context files, and sub-agent orchestration.
• Tool and function calling, plus designing tool interfaces a model can use reliably.
• Experience with multi-step agentic workflows including planning, tool selection, and error recovery.
• Context management, including deciding what belongs in the context window versus retrieval and managing cost and latency tradeoffs.
• Retrieval-augmented generation (RAG), including chunking strategy, embedding selection, hybrid search, reranking, and honest evaluation of retrieval quality.
• Structured output generation and schema validation.
• Model selection and routing across cost, latency, and capability tradeoffs.
• Strong Python skills to work confidently in a large existing codebase, including reading unfamiliar code, tracing bugs across modules, writing tests, and refactoring safely.
• API integration experience, with systems integration proficiency across REST APIs, OAuth 2.0 and token management, rate limiting, retries, pagination, and idempotency.
• Diagnostic judgment for issues that may fail silently, including proving that monitoring, data freshness, and data feeds are genuinely working.

Technologies

• TensorFlow, Hadoop, Spark, AWS, Talend, SQL
• OpenClaw (agent framework), retrieval system, vector search, model routing
• Microsoft Teams, Procore, Microsoft 365, Graph API, SharePoint
• OAuth 2.0, REST APIs
• Python, RAG (retrieval-augmented generation), embedding models, chunking strategy, hybrid search, reranking, schema validation
• Local inference alongside frontier API models
• OpenClaw (agent framework)

Preferred Qualifications

• Microsoft 365 and Graph API: mail, calendar, Teams, SharePoint, app registrations, and permission scopes.
• Vector databases in production, ideally Qdrant, and practical experience with embedding models.
• Local model hosting and serving: Ollama, vLLM, LiteLLM, and quantization tradeoffs.
• Mesh networking such as Tailscale, and basic reverse proxy configuration.
• PowerShell and Windows endpoint scripting.
• Evaluation frameworks for measuring LLM output quality.
• Front-end skills for internal dashboards and portals.
• Vision and OCR work for drawing analysis and scanned document extraction.

Compensation & Benefits

• $75,000 - $85,000 per year (compensation is commensurate with experience and discussed during the interview process).
• Full-time position with hybrid flexibility possible.
• Work location: In person (Boise, Idaho).
• Health insurance
• Dental insurance
• Vision insurance
• Paid time off

Equal Opportunity: MGI Inc. is an equal opportunity employer. All qualified applicants will receive consideration without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, or protected veteran status.