AI Automation Intern

Job Description

ABOUT ANTARES SOLUTIONS

\n

Antares Solutions is a Microsoft Solutions Partner with Advanced Specialisations in Copilot, AI on Azure, and Analytics on Azure. We help mid-market and enterprise clients build AI capability that actually ships; from proof-of-value agents through to production deployment.

\n

THE ROLE

\n

This is not a typical internship. You will work directly with our Leadership Team to design and build proof-of-value AI agents, starting with marketing automation and expanding into internal operations. Expect real deliverables from day one; not slide decks. This role is based in our Sydney office and you would be working minimum 3 days (Wednesday and Thursday compulsory in office days). This role is suited to penultimate or final year students at University.

\n

Initial focus areas include:

\n

\n
• Marketing automation - AI-assisted content generation, WordPress publishing, presentation decks, and landing pages
\n
• Internal operations - resource allocation, reporting, and pipeline analysis
\n

\n

WHAT WE’RE LOOKING FOR

\n

AI Tool Fluency

\n

\n
• Daily use of frontier LLMs - ChatGPT, Claude, Microsoft Copilot, Gemini; with a clear point of view on which to reach for and when
\n
• Habit of trying new model releases and feature drops in the same week, not reading about them later
  
\n

\n

Hands-On with Agent & Automation Platforms

\n

You’ve built something, however small with at least one of the following:

\n

\n
• Microsoft Copilot Studio or Power Automate / Power Platform
\n
• n8n, Make, or Zapier
\n
• OpenClaw or Hermes Agent (Nous Research)
\n
• LangChain, LangGraph, or CrewAI
\n
• Direct API integration with Anthropic, OpenAI, or Azure OpenAI
  
\n

\n

Prompt Engineering as a Working Skill

\n

\n
• Comfortable writing and iterating on system prompts, few-shot examples, and structured output prompts (JSON, XML)
\n
• Understands when to prompt vs. when to use RAG vs. when to fine-tune
\n
• Can explain why one prompt works better than another
  
\n

\n

Curiosity Signals We Want to See

\n

\n
• A portfolio of small AI projects, hackathon entries, side experiments, or open-source contributions — what you’ve built matters more than your transcript
\n
• Active engagement with the AI community, newsletters, LinkedIn, Discord, GitHub. You don’t need to be loud, just plugged in
\n
• A habit of testing multiple tools on the same problem and forming an opinion
\n
• Willingness to pull apart how something works rather than treating AI as a black box
\n

\n

BONUS EXPOSURE (NICE TO HAVE)

\n

\n
• Vector databases and RAG patterns - Azure AI Search, Pinecone, Chroma
\n
• LLM evaluation frameworks - LangSmith, Braintrust, or your own approach
\n
• MCP (Model Context Protocol) or agent-to-agent protocols
\n
• Computer-use or browser-automation agents
\n

\n

THE MINDSET TEST

\n

If your instinct when you see a new business problem is “I wonder which AI tool would solve this and how I’d test it” - you’ll fit here.

\n

If your instinct is “what’s the spec, I’ll build it” - this isn’t the right role.