Dongyu (Rain) Y.

AI Researcher

Pittsburgh, Pennsylvania, United States0 mo experience
AI EnabledAI ML Practitioner

Key Highlights

  • Published research at top-tier conferences.
  • Interned at Meta and TikTok, enhancing ML skills.
  • Expertise in Generative AI and Multimodal Learning.
Stackforce AI infers this person is a Machine Learning Engineer with a focus on AI and Computer Vision.

Contact

Skills

Core Skills

Machine LearningComputer VisionGenerative Ai

Other Skills

Vision Language Models (VLMs)Multimodal Large Language ModelsBenchmarkingLLMsChinese-CLIPBias DetectionVQA modelsContent ModerationMLLMsJailbreak TestingDe-hallucinationReliable Multi-modal ModelsDataset CurationData AnalysisVideo-LLMs

About

👋 Greetings! I'm Dongyu (Rain) Yao, a Master's student in Computer Vision (MSCV) at the Robotics Institute (RI), School of Computer Science (SCS), Carnegie Mellon University (CMU). 🔎 I'm actively seeking entry-level full-time opportunities in Machine Learning Engineer (MLE), Research Scientist, and SDE/SWE positions for spring 2026. 🔬My research interests broadly lie in Machine Learning, Computer Vision, NLP, and Generative AI (LLMs and MLLMs). My research endeavors paid off as my previous works were published at Top-tier conferences such as ACL (main), ICCV, ICASSP, CogSci (Oral), and NeurIPS. 🧑‍💻 I’m intoxicated with these cutting-edge technologies and their real-world applications. After honing my end-to-end modeling skills as a Machine Learning Engineer intern at TikTok (ByteDance), I deepened my engineering chops at Meta, building core ML infrastructure components and exploring new feature innovations. With hands-on model development insights and a solid engineering foundation on production-scale systems, I’m driven to push the boundaries of AI innovation even further. 🥁Apart from my professional experience as an enthusiastic researcher and engineer, I derive great enjoyment from music. I’ve had 18 years of performing experience with percussion instruments (e.g., snares, timpani, drum sets, marimba). I have served as the Chief percussionist at Wuhan University Symphony Orchestra and the drummer of the school music band AfterMathCandy. 🏀 I believe life thrives on movement. I have developed strong communication and cooperative skills as well as relaxing myself in the game of basketball (served as Shooting Guard in the department’s team of WHUCSE). I place significant importance on collective honor and team triumphs. 🤝 I’m always open to new opportunities. If you are interested in working with me, please feel free to drop an email to rain.dongyu.yao@gmail.com or raindy@cmu.edu

Experience

0 mo
Total Experience
--
Average Tenure
--
Current Experience

Meta

Software Engineer

May 2025Aug 2025 · 3 mos · Menlo Park, California, United States · On-site

  • SWE - ML intern @ Meta AI Infra Trainer team (AI Workflows)
  • Completed with a rating of Exceed Expectations (EE)

Carnegie mellon university

Research Assistant

Jan 2025Jan 2026 · 1 yr · Pittsburgh, Pennsylvania, United States · On-site

  • · Conducting research on de-hallucination for Vision Language Models (VLMs) that comprehend multi-image sequences and videos.
  • · Supervised by Prof. Katia Sycara (https://www.ri.cmu.edu/ri-faculty/katia-sycara/) and Dr. Yaqi Xie (https://yaqi-xie.me/)
Vision Language Models (VLMs)Multimodal Large Language ModelsMachine LearningComputer Vision

Massachusetts institute of technology

Research Assistant

Nov 2024Mar 2025 · 4 mos · Remote

  • · Research Assistant at MIT and HKUST.
  • · Benchmarking Vision Language Models on visually linking explicit matching cues across multi-image sequences and videos.
  • · Supervised by Prof. Paul Pu Liang [https://pliang279.github.io/] and Dr. Yi R. (May) Fung [https://mayrfung.github.io/] (now AP at HKUST).
BenchmarkingVision Language Models (VLMs)Machine LearningComputer Vision

Tiktok

Machine Learning Engineer

Apr 2024Jul 2024 · 3 mos · Haidian District, Beijing, China · On-site

  • Machine Learning Engineer Intern @ Interactive Entertainment Service - Quality Assurance Team
  • (Co-first Inventor in Patent) Engineered an automated pipeline using LLMs and Chinese-CLIP to detect bias in text-to-image generation for TikTok and CapCut, introducing a novel open-case approach better suited for industrial deployment.
  • Developed a content moderation system using VQA models with custom prompts to detect AIGC-generated nudity risks and child-inappropriate content for CapCut’s Dreamina, achieving greater automation in the review process.
  • Proposed novel jailbreak methods for MLLMs by embedding text in images and leveraging discrepancies in tuning between text and image inputs to disrupt attention during alignment, successfully breaching internal models on Coze.
  • Collaborated with team members on zero-shot evaluations of upcoming multimodal models on Byteval (internal evaluation platform), and proposed adversarial robustness assessments based on previous AI security expertise.
Multimodal Large Language ModelsGenerative AIMachine LearningComputer Vision

Education

Carnegie Mellon University

Master of Science - MS — Computer Vision

Aug 2024Dec 2025

Wuhan University

Bachelor of Engineering - BE — Cybersecurity

Jan 2020Jan 2024

Stackforce found 100+ more professionals with Machine Learning & Computer Vision

Explore similar profiles based on matching skills and experience