Qingyuan Lu

Appearance

About me ​

I graduated from MIT in June 2023 with a B.S in Computation and Cognition, an interdisciplinary degree which combines elements of Computer Science and Brain and Cognitive Sciences. I have been coding in various languages since middle school, and I am especially passionate about using computer science to build and study intelligence. I am invested in finding ways to make sure algorithms and AI models are safe and reliable as they have an increasing impact on people's daily lives.

As a recent graduate, I am currently looking for jobs in Software Engineering/Development to build practical skills in working on complicated software projects deployed in the real world, with a long-term goal of working in more research-related roles in the future.

Note: I changed my first name late 2022 so some of my previous work is credited under a different name.

Work experiences ​

Center for Human-Compatible AI: Intern (Dec 2023-Jun 2024)

  • Full-time internship
  • Created StrongREJECT, a novel benchmark for jailbreaking attacks against LLMs consisting of 346 prompts for evaluating the potential of jailbreak attacks to enable malicious actors to cause harm and an autograder with SOTA performance at matching fine-grained human evaluations of jailbreak quality
  • Implemented tests of jailbreak effects on intelligence using Massive Multitask Language Understanding benchmark.
  • Fine-tuned LLM to develop cost-effective autograding method which can be run on a laptop CPU.

Concordia Consulting: AI Safety Affiliate (Jun 2023-Nov 2023)

  • Project-based, ~10hrs/month
  • Translated and revised texts in English and Chinese about AI safety, including Chinese-language social media posts about technical AI safety for general audiences and translations of Chinese public figures' comments about AI safety for English-speaking audiences.

Seven Seas Entertainment: Freelance Translator (Jul 2021-Present)

  • Project-based freelance position.
  • Translated 10 novels from Chinese to English for North American publication, including 3 New York Times best-selling volumes.

Concordia Consulting: AI Governance Affiliate (Nov 2022-Apr 2023)

  • Project-based, ~10hrs/month
  • Translated and revised texts in English and Chinese about AI safety and governance, including socia media posts and UN submissions.

Center for Human-Compatible AI: Intern (Jun 2022-Nov 2022)

  • Full-time during summer, part-time ~10hrs/week during semester
  • Developed a novel mathematical framework for ethically compliant autonomous systems capable of operating in partially observable domains.
  • Implemented simulated environments and solution method in Python and CPLEX.
  • Ran experiments on Linux server using Docker containers.
  • Drafted paper as first author, accepted for ICRA 2024.

MIT Interactive Robotics Group: Undergraduate Researcher (Jan 2022-May 2022)

  • Part time ~10hrs/week
  • Contributed features to a PyTorch multiagent RL code base to study the effects of a novel method for communication between agents.
  • Created visualization code to illustrate communication behavior.

MIT Experimental Phonetics Lab: Undergraduate Researcher (Nov 2019-Mar 2020, Feb 2021-Aug 2021)

  • Full-time during summer, part-time ~10hrs/week during semester
  • Wrote software for offline and online phonetics experiments using PsychoPy (Visual interface + Python) and PsychoJS (JavaScript) to deploy on Pavlovia online platform and collect hundreds of responses within days.
  • Wrote Python scripts to process thousands of datapoints and analyze results.
  • Assisted with running experiments on adult participants in-person.
  • Recorded and processed speech samples for experiment materials using Praat.

MIT Concourse: Classical Mechanics Teaching Assistant (Sep 2020-Dec 2020)

  • Part-time ~12hrs/week
  • Led problem-solving recitations and small group meetings to support student success in introductory physics course during remote learning. Collaborated with other TAs to grade homework.
  • Tutored 1 student 1-on-1 weekly to provide extra assistance with course material.

AbbVie: Information Research Intern (Jun 2020-Aug 2020)

  • Full-time
  • Conducted independent research to study a technique to fine-tune machine translation models in low-resource domains using PyTorch. Ran experiments on Linux server.

Skills ​

Programming languages

  • Python (primary)
  • JavaScript
  • Java
  • C++
  • Julia
  • MatLab
  • HTML
  • CSS

Software and packages:

  • GitHub
  • Docker
  • Pytorch
  • TensorFlow
  • Google Apps Script
  • PsychoPy
  • Praat
  • WordPress
  • Joomla

Applications:

  • Machine learning algorithms
  • Natural language processing
  • Computer vision
  • Probabilistic programming
  • Bayesian approaches

Math and algorithms:

  • Algorithm design and runtime analysis
  • Data structures
  • Linear algebra
  • Probability and statistics
  • Differential equations
  • Signal processing

Miscellaneous:

  • Linux
  • Mandarin Chinese (fluent)