AI Trainer

Metis Strategy

1d•$140,000 - $220,000

About The Position

Design and deliver hands-on AI training for enterprise clients. Training sessions are mostly virtual, run in 3–4 hour blocks across global time zones, and serve audiences ranging from executives to software engineers. Builder cohorts work directly in labs on skills, agents, MCP servers, coding agents, and code review, while leader and enabler cohorts focus on everyday workflows. The AI Trainer co-facilitates with other Metis trainers and tailors content to each client’s industry, tools, and security requirements. The role owns keeping training content current, updating materials before each cohort as models, tools, and pricing change quarterly. The position requires production experience with LLM or agentic systems, breadth across major AI platforms and tools, and the ability to teach and facilitate live technical sessions. The trainer must be able to troubleshoot enterprise setup issues, communicate complex concepts to varied audiences, manage live demonstrations calmly, and respond thoughtfully to skepticism and job-displacement concerns. Successful candidates are motivated to help clients achieve strategic and operational excellence while continuously developing their skills and engaging with emerging intellectual challenges. The position is eligible for bonus compensation; compensation and benefits are competitive and commensurate with qualifications and relevant work experience.

Requirements

10+ years in software development, data/ML engineering, architecture, or technical product work, with recent practical experience building with LLMs.
Has shipped LLM or agentic systems to production and can show examples.
Fluent in agentic coding harnesses, including context files (CLAUDE.md / AGENTS.md), skills, subagents, hooks, MCP, and plugins.
Has built a custom MCP server and an agent end to end.
Understands prompting, RAG, fine-tuning, and self-hosting, and can explain when to use each.
Understands agent security, including prompt injection, least-privilege tools, human-in-the-loop, and guardrails outside the model.
Can explain Tokenomics and cost control to engineering and finance audiences.
Hands-on experience with Anthropic (Claude, Claude Code, Cowork), OpenAI (ChatGPT, Codex, SDKs), Microsoft (M365 Copilot, Copilot Studio, Copilot Cowork, GitHub Copilot), and open-weight models.
Working familiarity with Google (Gemini, Vertex), AWS (Bedrock), Cursor, and major agent frameworks (LangGraph, CrewAI, MS Agent Framework).
Has taught live, technical, hands-on sessions to groups, in person or virtually.
Can pace a four-hour virtual session using polls, breakouts, chat, energy management, and a clear break structure.
Can troubleshoot setup problems in locked-down enterprise environments, including Windows, WSL2, proxies, restricted installs, SSO, and licensing.
Can explain complex concepts plainly to different audiences and stay calm when a live demo misbehaves.
Can read a remote room and adjust to participants’ varying levels of understanding and engagement.
Can address skepticism and job-displacement anxiety directly without dismissing them.
Comfortable saying “I don’t know, let’s test it.”

Nice To Haves

Consulting or client-facing delivery experience, including working under a partner brand and co-facilitating.
Governance and compliance fluency, including responsible AI, data classification, the EU AI Act, and data residency.
Evals and observability experience, including building test sets, guardrails, and monitoring for LLM applications.
Instructional design experience, including building labs, measuring learning with pre/post assessments, and reinforcing learning through office hours or 30/60/90-day follow-through.
Change management or AI adoption experience, including connecting training to business outcomes and ROI.
Domain experience in a regulated or industrial sector, such as manufacturing, IT/OT, healthcare, or financial services.
Experience authoring skills, plugins, or reusable templates that a client team can maintain.
Experience teaching across cultures and to non-native English audiences, for example across India, EMEA, and North America in the same program.
Public thought leadership through writing, talks, or open-source work.
Self-hosted inference and fine-tuning experience, such as vLLM, Ollama, or LoRA/QLoRA.

Responsibilities

Design and deliver hands-on AI training for enterprise clients.
Facilitate mostly virtual sessions in 3–4 hour blocks across global time zones for audiences ranging from executives to software engineers.
Lead builder cohort labs covering skills, agents, MCP servers, coding agents, and code review, and support leader and enabler cohorts focused on everyday workflows.
Co-facilitate sessions with other Metis trainers.
Tailor content to each client’s industry, tools, and security requirements.
Keep training content current and update materials before each cohort as models, tools, and pricing change.
Troubleshoot setup issues in locked-down enterprise environments, including Windows, WSL2, proxies, restricted installs, SSO, and licensing.
Explain complex concepts plainly to different audiences, manage virtual session pacing, and remain calm when live demonstrations misbehave.
Read the room and adjust to participants’ needs, address skepticism and job-displacement anxiety, and acknowledge when an answer requires testing.