Application Guide

How to Apply for Technical Advisor, Fine-Tuning and Evals

at Mycelium

🏢 About Mycelium

Mycelium is a unique organization dedicated to advancing AI systems that consider the welfare of all sentient beings. By building benchmarks and tools, they are at the forefront of a nascent field that integrates ethics and AI. Working here offers the chance to contribute to a mission-driven company that's shaping the future of AI alignment.

About This Role

This role involves advising on fine-tuning and evaluation design to instill animal welfare values in open-weight models. You'll guide experimental design, review synthetic data pipelines, and create robust evaluations. Your work will directly impact how AI models treat animals, making it both intellectually challenging and ethically significant.

💡 A Day in the Life

A typical day might involve reviewing a synthetic data generation pipeline, advising on the design of a new evaluation benchmark, and collaborating with researchers on fine-tuning experiments. You'd likely spend time writing feedback documents, participating in virtual meetings, and staying updated on the latest alignment research. The remote, part-time nature allows for flexibility, but you'd be deeply engaged in the technical and ethical aspects of the work.

🎯 Who Mycelium Is Looking For

  • Experienced in fine-tuning large language models, with a deep understanding of techniques like instruction tuning, RLHF, and parameter-efficient methods.
  • Skilled in experimental design, including control arms, ablations, and statistical validation, to ensure rigorous and reliable results.
  • Proficient in building and evaluating synthetic data pipelines, with an eye for quality and suitability for specific value alignment.
  • Familiar with evaluation metrics and methodologies that measure out-of-distribution generalization and avoid data leakage.
  • Knowledgeable about AI alignment literature, especially value alignment and ethics, with a passion for animal welfare.

📝 Tips for Applying to Mycelium

1

Highlight specific projects where you've fine-tuned open-weight models (like Llama or Mistral) for value alignment, and mention the metrics you used to measure success.

2

Describe a time you designed an evaluation that tested out-of-distribution generalization; explain the methodology and how you prevented data leakage.

3

Show your familiarity with synthetic data generation by discussing how you've created or reviewed datasets for quality and bias.

4

Mention any experience with animal welfare or ethics in AI, even if it's a personal interest or side project, to demonstrate genuine alignment with Mycelium's mission.

5

Tailor your resume to emphasize experimental design and statistical rigor, as these are core to the role.

✉️ What to Emphasize in Your Cover Letter

["Emphasize your passion for animal welfare and how it aligns with Mycelium's mission to consider all sentient beings in AI.", 'Detail your hands-on experience with fine-tuning and evaluation, providing concrete examples of projects and outcomes.', 'Show your ability to think critically about experimental design, including how you handle confounding variables and ensure validity.', 'Mention any familiarity with open-weight models and the open-source AI community, as this is central to the role.']

Generate Cover Letter →

🔍 Research Before Applying

To stand out, make sure you've researched:

  • Read Mycelium's website and any published papers or blog posts to understand their current benchmarks and tools.
  • Research the concept of 'welfare range' and how it's used in animal welfare science, as it may inform evaluation design.
  • Look into recent work on value alignment in open-source models, such as Anthropic's or OpenAI's alignment research, to see where Mycelium might fit.
  • Explore the debate around moral consideration for animals in AI, including critiques and defenses, to be prepared for nuanced discussions.
Visit Mycelium's Website →

💬 Prepare for These Interview Topics

Based on this role, you may be asked about:

1 How would you design an experiment to evaluate whether a fine-tuned model genuinely cares about animal welfare versus just following instructions?
2 What are the potential pitfalls in using synthetic data for value alignment, and how would you mitigate them?
3 Describe a time you had to balance multiple objectives in an alignment project; how did you prioritize?
4 How do you ensure that evaluations measure out-of-distribution generalization without leaking training data?
5 What are your thoughts on the current state of AI alignment literature, and where does animal welfare fit in?
Practice Interview Questions →

⚠️ Common Mistakes to Avoid

  • Don't submit a generic application that doesn't mention animal welfare or Mycelium's mission; it's a core part of the role.
  • Avoid being vague about your experience; provide specific examples and metrics to demonstrate your expertise.
  • Don't overlook the importance of experimental design; even if you're strong in ML, showing you can think rigorously about experiments is crucial.

📅 Application Timeline

⏰ Deadline: September 11, 2026

We recommend applying at least a few days early to avoid last-minute technical issues.

Typical hiring timeline:

1

Application Review

1-2 weeks

2

Initial Screening

Phone call or written assessment

3

Interviews

1-2 rounds, usually virtual

Offer

Congratulations!

Ready to Apply?

Good luck with your application to Mycelium!