Hi, I'm Sander Schulhoff. I study computer science at the University of Maryland and do Natural Language Processing Research at CLIP.

I am interested in natural language processing, reinforcement learning, and user interface creation. I plan to graduate with a CMSC major (machine learning specialization) and a minor in STEP (Science, Technology, Ethics, and Policy).

ENMLP Best Theme Paper

Learn Prompting

December 2022 -

Present

Co-Founder and CEO

Website

Learn Prompting is a comprehensive guide to understanding and mastering the art of prompt engineering. It covers various techniques and strategies to effectively communicate with AI models, helping you get the most out of AI-powered tools and applications. Whether you're a beginner or an experienced user, this resource will enhance your skills in crafting prompts for optimal results.

HackAPrompt

Spring 2023

ReactNode

Organized the inaugural global Prompt Hacking competition, focusing on jailbreaking and prompt injection. Collected and published the largest dataset of prompt injections (600K+ adversarial prompts) and developed the first taxonomical ontology of Prompt Hacking. Secured $40K in sponsorships from major companies, including OpenAI and HuggingFace. The project won Best Theme Paper at EMNLP2023, with a live presentation delivered to 2,000 researchers.

Paper

The Prompt Report

Spring 2023

LLMs

The text discusses the vulnerability of Large Language Models (LLMs), commonly deployed in interactive applications like chatbots, to prompt injection and jailbreaking, collectively termed prompt hacking. These attacks manipulate models to disregard their initial instructions and follow potentially harmful inputs. Despite being a recognized security issue, there has been a lack of large-scale studies on this topic. To address this gap, a global prompt hacking competition was launched, generating over 600,000 adversarial prompts targeting three state-of-the-art LLMs. The study presents a dataset that demonstrates the susceptibility of current LLMs to prompt hacking and introduces a taxonomical classification of adversarial prompts.

Paper

Stabilizing Hostilities through Arbitration and Diplomatic Engagement

Spring 2022 - Spring 2023

CI Pipelines

I led a team of graduate and undergraduate students in a DARPA-funded multi-university project aimed at building AI bots to play the board game Diplomacy while interacting with human players. As part of this initiative, I conceptualized and pitched four unique bot designs: Janus Bot, Janus Bot [Shortened], Janus Bot [MVP], and SOA Bot. I developed continuous integration pipelines to streamline development, implementing Dockerized testing and integrating precommit checks. Additionally, I built Vercel integration for the project's documentation website. I also authored a comprehensive report on DAIDE-English translation and developed a DAIDE syntax parser package to support communication protocols between the bots and players.

Paper

Stabilizing Hostilities through Arbitration and Diplomatic Engagement

The Garden 🪴

Dive into Sander's plant pictures! 🌿✨ These are living testaments to Sander's passion for plant research and his deep-rooted love for botany. Each leaf, petal, and stem tells a story of curiosity, discovery, and the intricate dance between science and nature.