LLM Engineer (Remote, Nelspruit)

Blacksmith Agency LLC; Mbombela (Nelspruit); Remote; ZAR 900,000 - 1,300,000; Full time

Job summary

Blacksmith Agency is seeking an LLM Engineer to own and ship our internal automation roadmap, focusing on production systems rather than research. You will work with a US team on practical deployments and guardrails.

The role includes building agents with tool use, integrations with WordPress, Figma, Google Workspace, Slack, and internal MCP servers, and delivering internal tools that cut hours across delivery and operations.

Qualifications

• 2+ years building production software in Python or TypeScript/Node.js.
• At least one LLM or agent system built in production with real users.
• Hands-on tool use, evals and failure handling: validating model output, retries, human approval.
• REST APIs, webhooks and cloud deployment.
• Strong spoken English; able to discuss technical decisions on live calls with a US team.

Responsibilities

• Own and ship internal automation roadmap, production LLM and agent systems.
• Build agents and workflows with tool use and function calling via APIs.
• Create integrations with WordPress, Figma, Google Workspace, Slack and internal MCP servers.
• Develop internal tools that reduce hours in delivery, sales ops, and project management.

Skills

Python/TypeScript

LLM/agent systems

REST APIs

English communication

Tools

WordPress

Figma

Google Workspace

Slack

Job description

About The Role

Blacksmith Agency is building AI into every part of how we work. As LLM Engineer at Blacksmith, you own and ship our internal automation roadmap: production LLM and agent systems that real people on our team use every day. This is a builder role, not research. You will not train models.

Blacksmith Agency

• Engineering
• Full time, your only job
• Remote (Latin America and South Africa)

What You Will Build

• Agents and workflows with tool use and function calling on Anthropic and OpenAI, with a framework or plain APIs.
• Evals that prove a change made things better, and guardrails that stop a model taking a wrong or unsafe action.
• Integrations connecting WordPress, Figma, Google Workspace, Slack and our internal systems, including MCP servers.
• Internal tools that take measurable hours off delivery, sales operations and project management.

What You Bring

• 2+ years building production software in Python or TypeScript / Node.js.
• At least one LLM or agent system you built yourself that ran in production with real users.
• Hands-on tool use, evals and failure handling: validating model output, retries, human approval.
• REST APIs, webhooks and cloud deployment.
• Strong spoken English. You will talk through technical decisions on live calls with a US team.

Not a fit if

• You freelance or contract for other clients and would keep them alongside this job.
• Your LLM work is demos, tutorials or chat prompting, or your focus is model training and research.