Hugging Face Just Released an AI Intern That Might Take Your Junior Dev Job
Hugging Face is rolling out a new AI agent, ML Intern, that handles research, coding, and model deployment autonomously. It is a bold attempt to turn a chat box into a tireless, coffee-free junior engineer that actually reads the documentation.
ML Intern operates on a research-first philosophy, forcing the agent to scan academic papers and library documentation before writing a single line of code. It manages the full lifecycle of a machine learning project, from selecting datasets and model architectures to debugging code and launching cloud-based training sessions.
The system uses an agentic loop to evaluate its own progress, utilizing up to 300 integrations per task. If it gets stuck in a logic loop, it features a built-in detection system to reset its approach. Upon completion, it can deploy a model repository, a Hugging Face Spaces demo, or a Trackio dashboard for monitoring.
Under the hood, the agent runs on GLM-5.2 via Novita by default, though users can swap it for GPT, Claude, Kimi, MiniMax, or DeepSeek. For those who prefer keeping their secrets close to home, it supports local connections via Ollama, vLLM, LM Studio, or llama.cpp, and the source code is now live on GitHub.
It is cute that the tech industry keeps building tools to automate the very people they just finished laying off. Eventually, the only thing left for humans will be to hold meetings about the AI agents that are currently having meetings with other AI agents to decide which model to hallucinate next.
Source: Hugging Face
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.