Specialist-model research desk

Expert Q&A

How should the interview be commissioned and published?

Frontier Development research deskSection 06 · Page 13 of 17Last updated 26 October 2026

Send the scope, audience and questions in advance. Request disclosure of employer, investments, paid advisory work and supplier relationships relevant to the discussion. Obtain consent for recording and a clear quotation-approval process; no outreach or recording has been performed for this pack.

During the interview, distinguish firsthand experience, published evidence and opinion. Ask for links and dates when a number or provider claim appears. Afterward, verify factual claims independently, preserve the meaning of quotations and label unanswered questions. Do not transform a cautious answer into a categorical recommendation.

The publishable article should lead with the buying decision and the evidence threshold, then present the expert’s actual answer, a source-linked desk note and an unresolved question. Keep the blank quote fields unpublished until an authorised contribution exists.

TRACE THE EVIDENCE

Section source library

This page contains no direct external citations. These are the section’s references; they do not independently validate the desk’s proposed scoring or methods.

  1. QLoRA: Efficient Finetuning of Quantized LLMs — Dettmers et al., 23 May 2023 →arxiv.org
  2. A Field Guide to Rapidly Improving AI Products — Hamel Husain →hamel.dev
  3. Why AI Compute Comes in Rectangles — Erik Bernhardsson, Latent Space, 22 October 2026 →latent.space
  4. SB26-189 Automated Decision-Making Technology — Colorado General Assembly →leg.colorado.gov
  5. Demystifying evals for AI agents — Anthropic, 9 January 2026 →www.anthropic.com
  6. LoRA Learns Less and Forgets Less — Biderman et al., revision 20 September 2024 →arxiv.org
  7. The Inference Engineering Masterclass — Philip Kiely and Ali Taha, Latent Space, 3 August 2026 →latent.space
  8. GroupKFold — scikit-learn documentation →scikit-learn.org
  9. RouteLLM: Learning to Route LLMs with Preference Data — Ong et al., revision 23 February 2025 →arxiv.org
  10. Llama 3.3 Community License Agreement — Meta →github.com