I'm Will Kalikman, a machine learning research engineer at Beacon Holdings. Before that I was a Master's student in computer science at ETH Zurich. My thesis was on gradient leakage attacks with the SRI lab and INSAIT in the secure and trustworthy AI group.

Before starting my thesis, I spent most of 2025 researching optimization-based adversarial attacks on language models. My research interests are in search and what it can reveal about how machine learning models work.

I received my undergraduate degree from Dartmouth College (BS CS & BA Philosophy) where I spent a lot of time rowing.

All pictures on the site taken by yours truly.

You can reach me at: w[lastname]atgmail dot com.

• Augmenting Text To Increase Translation Difficulty May 4, 2026

  A paper on augmenting source text to make machine translation harder.

• LLM Attacks: GCG, Benevolent Fine-Tuning, and Backdoor Embedding April 19, 2026

  My first research project in adversarial ML.

• Appreciating the Burden of Creation February 15, 2026

  Why I made this site.

• Combining Real Counterfactuals with Saliency Maps September 11, 2025

  On building interpretable ML tools at Coinbase.

• Understanding Is an Illusory Goal May 15, 2025

  Old notes on what we actually want when we ask for explanations of ML systems.