Lukas Xue

About

I build foundation-model systems across post-training, alignment, and inference. My work sits between research and production: turning model-customization ideas into scalable training and serving systems for enterprise GenAI workloads.

I am currently an MLE II in the AWS Generative AI Innovation Center and a recurring contributor/code owner in AWS open source.

Post-training Alignment Inference optimization Distributed training Model steering

Experience

Feb 2023 – Present

Machine Learning Engineer

Amazon Web Services

ML and systems engineering across AWS Generative AI Innovation Center, Amazon Bedrock, SageMaker, and Managed Streaming for Apache Kafka. Promoted to SDE II in April 2025.

May 2025 – Present

MLE · Custom Model Optimization (CMO)

AWS Generative AI Innovation Center

Work on LLM/VLM post-training, distributed training, inference optimization, and agentic-system components for enterprise GenAI customers.

Aug 2024 – May 2025

SDE → SDE II · Bedrock Model Customization

Built model-customization infrastructure for fine-tuning, distillation, and multi-tenant adapter serving in Amazon Bedrock.

Apr 2024 – Aug 2024

SDE · SageMaker AutoPilot & Canvas

AutoML for tabular data, time-series forecasting, and language model fine-tuning. Helped make ML accessible to non-practitioners through Canvas's no-code interface.

Feb 2023 – Apr 2024

SDE · Managed Streaming for Apache Kafka (MSK)

Worked on Kafka infrastructure, AWS Graviton support, and managed streaming systems at AWS scale.

May 2022 – Aug 2022

SDE Intern, MSK

Amazon Web Services

Built real-time fault-tolerant data pipelines and streaming infrastructure. First exposure to distributed systems at AWS scale.

Summer 2021

Full Stack Intern

A.P. Moller – Maersk

Modernized Maersk's shipping logistics platform. Built features for container tracking and shipment management across global supply chain operations.

Education

M.S. Information Studies

Trine University

Aug 2025 – Present · Part-time

GPA: 4.0/4.0. Focus on data systems, information architecture, and applied research.

B.S. Computer Science

Emory University

2021 – Dec 2022

Minor in Applied Mathematics. GPA: 3.938/4.000. Coursework in algorithms, systems, ML, and mathematical foundations.

Student Researcher — Data mining and graph neural networks.
Teaching Assistant — CS334 Machine Learning, CS326 Analysis of Algorithms.

Associate of Arts

Oxford College of Emory University

2019 – 2021

Liberal arts foundation before transitioning to Atlanta for upper-division CS coursework.

Publications

Inference-Time Model Steering via Predictive-State Intervention: A Survey · R. Xue, R. Wang, Y. Wang, Y. Cui, Y. Qian, P. Vaddamanu, H. Song, H. Marlowe · Preprint, 2026
From Errors to Rules: Iterative Prompt Optimization for Text Classification · Y. Cui, R. Xue, Y. Zhang, M. Prasad · Preprint, 2026
PRISM: Position-encoded Regressive Inverse Spectral Model for Multilayer Thin-Film Design · R. Wang*, R. Xue*, B. Chen, H. Wu · ICML Workshop on AI for Physics, 2026
AME-TS: Anchored Mixture-of-Experts for Time Series Forecasting · R. Wang, R. Xue, R. Razi, H. Song, H.R. Marlowe · ICML Workshop on Structured Foundation Models, 2026
Deep Graph Representation Learning and Optimization for Influence Maximization · C. Ling, J. Jiang, J. Wang, M.T. Thai, R. Xue, J. Song, M. Qiu, L. Zhao · International Conference on Machine Learning (ICML), 2023

* Equal contribution

Projects

awesome-inference-time-steering — Curated paper collection on inference-time model steering via internal interventions. Paper
PRISM — Autoregressive transformer for inverse thin-film optical design. Playground · Paper
nanoLLaDA — Minimal masked diffusion language model (~500 lines). Pretraining & generation for LLaDA.
awslabs/mcp — Code owner and recurring contributor to AWS's official Model Context Protocol repository, including serverless MCP infrastructure.
OpenAcademiaRating — Open-source, privacy-first professor rating platform focused on graduate advising and mentorship outcomes.
AI Paper Digest — Daily top 5 AI/ML papers ranked by community upvotes, with generated summaries and detailed modal views.

Lukas (Renhao) Xue

About

Experience

Education

Publications

Projects

Certifications