Technical writer for AI research
Your research is good.
I make sure people finish reading it.
I’ve written papers, reports, and explainers with research teams at Anthropic, Epoch AI, Foresight Institute, and Palisade Research. I was a researcher first, at Google Brain and Mila, so I can talk to researchers in their own terms.
Two ways to work with me
What I do
Writing and editing
You have results, plots, and a draft nobody outside the team can follow. I turn it into a paper, a report, or a thread that lands and stays true to what you found. Alignment Faking, SHADE-Arena, the Secure AI Tech Tree, and the robot-dog thread.
See the work →Research coaching
For researchers with results and no draft, or ideas and no plan. We work out what the claim is, what to run next, and what to drop, and I keep the project moving until it’s done. For the tangent-prone, and the ones with a thousand plots.
Email me →Selected work
I wrote these
Nebius Science · 2026
The Web Has a New Reader
Search was built for people. Research agents read differently. How Tavily rebuilt it for them.
Palisade Research · 2026
Shutdown resistance in LLM-controlled robots
A robot dog, a shutdown order, and a model that kept going. 1.3 million views.
Foresight Institute · 2025
Secure AI Tech Tree
The whole field of secure AI, mapped as one tree. Concept, content, and launch.
Bezos Earth Fund · 2024
AI for Climate and Nature
Where machine learning helps on climate, and where it doesn’t. Workshop report.
Substack · 2026
How Much of Your Emotional Life Runs Through a Chatbot?
Eight long interviews about people and their AIs, and the five tensions that kept coming up.
I edited these
Anthropic · 2024
Alignment Faking in Large Language Models
The paper that caught a frontier model playing along during training. Editing and figures.
Foresight Institute · 2026
Open Science Needs Open Compute
Allison Duettmann’s essay on why independent science depends on compute, not just funding. Editorial rewrite.
Epoch AI · 2026
The Plunging Price of Thought
The cost of a given level of AI performance has fallen about 47% a quarter since 2023, faster than any earlier technology. Editing.
The short version
How I got here
I trained as a mathematician, then as an AI researcher. At Google Brain I coordinated a team of computer scientists and biologists using interpretability tools to read the epigenome. My master’s thesis became an ICLR paper. I was a decent researcher. I was a better explainer.
In 2023 I left my PhD and started writing full time. Anthropic brought me in to edit and draw figures for papers like Alignment Faking. Palisade Research had me sit with the researchers, read the eval logs, and turn hacking demos into threads. Foresight Institute asked me to map the whole field of secure AI, from concept to launch.
Today most of my time goes to papers with Anthropic, with some work for Epoch AI. On Substack I write about the part that never makes it into the paper: what living with these systems does to us.
I live in rural Italy. I teach acroyoga. I answer email.
In the 10 years I have been involved in hiring contractors for various technical writing at Foresight, Linda has been the best writer I’ve worked with. […] Even if the topic is technically complex, and outside her domain knowledge, she communicates concepts correctly, effectively and with an interesting lens.
Allison Duettmann
President, Foresight Institute
The writing itself has been accurate and well-framed; she engages with the underlying research rather than just packaging it. […] Not just as a writer but as someone who thinks carefully about what the evaluations show.
Jeffrey Ladish
Executive Director, Palisade Research
Contact
Send me a draft, a paper in progress, or a rough idea of what needs writing.
I’ll tell you within a couple of days whether I can help.
Rates start at €150 an hour. Most engagements run 10 to 35 hours a week for a few months.
Essays go out on Substack when there’s something to say. Subscribe.