About OpenTrain

OpenTrain AI is the hiring and contracting organization for this role and the #1 platform for finding and building careers in AI training and data labeling. OpenTrain helps people discover opportunities, build an AI-work profile, and apply in minutes, with free account creation.

This contract places your software engineering and machine learning expertise at the center of how an AI coding model learns specialized audio development practices.

• Worldwide remote opportunity
• Part-time contractor engagement
• English-language project
• Apply through OpenTrain

About AI Training Work

AI training is the human side of building modern artificial intelligence. Specialists prepare examples, write and review model responses, and evaluate technical content so models can produce more useful and reliable results.

In this project, your work will combine code curation, instruction-response writing, quality review, and model fine-tuning. It is a hands-on opportunity to shape a coding model for a specialized programming domain.

• Work remotely with a computer and internet connection
• Use expert knowledge to prepare high-value training examples
• Contribute to cutting-edge generative AI development
• Choose a flexible workload around a 20+ hour weekly commitment

The Role

OpenTrain AI is seeking an expert ML engineer to build a supervised fine-tuning dataset for a JUCE/C++ audio DSP coding model and run QLoRA fine-tuning on Qwen3-Coder. The target dataset contains 3,000 to 5,000 examples covering practical audio plugin development and real-time digital signal processing.

The project focuses on extracting and transforming high-quality technical material from open-source repositories and educational resources. A comprehensive resource document with repository URLs, blog links, textbook references, and a clone script will be provided.

• Fixed-price compensation: $1,000 USD
• Time requirement: 20+ hours per week
• Experience level: Expert
• Employment type: Contractor and part-time
• Work location: Worldwide

What You'll Do

You will create the source material and training examples needed for a specialized coding model. The work includes both technical data preparation and model-training execution.

• Extract DSP-relevant C++ functions from 40+ open-source GitHub repositories, including Surge, ChowDSP, Airwindows, Vital, and the JUCE framework
• Generate high-quality instruction-response pairs with LLM-assisted pipelines such as Bespoke Curator or Distilabel using Claude or GPT-4
• Convert blog posts, tutorials, forum Q&A, and free textbook content into clean ChatML-formatted training examples
• Run a second LLM quality-filtering pass and deduplicate the resulting examples
• Run QLoRA fine-tuning on Qwen3-Coder using Unsloth
• Build a target dataset of 3,000 to 5,000 examples
• Cover processBlock, AudioBuffer, juce_dsp filters, oscillators, delay lines, reverb, virtual analog modeling, plugin architecture, and real-time DSP best practices

Required Expertise

This is an expert-level role requiring the ability to work across machine learning workflows, C++ code, and audio DSP concepts. The project specifically calls for specialized experience rather than general annotation alone.

• Expert-level ML engineering experience
• Strong C++ programming capability for coding dataset work
• Knowledge of digital signal processing and C++ audio plugin development
• Ability to curate, transform, filter, and deduplicate technical training data
• Experience running or implementing fine-tuning workflows is relevant to the project scope
• English-language working ability

Helpful Background

Previous experience with the JUCE framework would be especially helpful. Candidates are also encouraged to share examples of audio projects they have worked on, if available, so their relevant experience can be reviewed.

• Prior JUCE experience is helpful
• Examples of audio projects are helpful
• Experience with C++ audio plugin development is directly relevant
• Familiarity with DSP implementation concepts is directly relevant

How to Apply

Create a free OpenTrain account and apply through OpenTrain AI. In your application, highlight your ML engineering, C++, JUCE, and audio DSP experience, and include relevant audio project examples if you have them.

The selected contractor will receive the project resource document containing repository URLs, educational links, references, and the clone script needed to begin the curation and fine-tuning work.

• Apply remotely from anywhere in the world
• Showcase relevant JUCE or audio plugin work
• Mention experience with dataset curation or fine-tuning workflows
• Be prepared to commit 20+ hours per week