Songting (Michael) Wang

I’m a Master’s and Bachelor’s student in Electrical and Computer Engineering at Carnegie Mellon University, with a minor in Computer Science.

My interests are in ML systems, compilers, and distributed infrastructure — building software that makes ML workloads faster and more efficient.

Research: I’m a Graduate Researcher with Google CoreML × CMU Catalyst, building a compiler to lower Mirage-generated computation graphs into efficient Google TPU kernels (advised by Prof. Zhihao Jia). I’m also working on Mirage Persistent Kernel, the first compiler and runtime system that automatically transforms multi-GPU model inference into a single high-performance megakernel.

Separately, I work with Prof. Vyas Sekar at the CyLab Security & Privacy Institute on data systems — I was an invited speaker at Current 2025 (world’s largest Data Streaming conference), open-sourced FlinkSketch — a high-performance library of probabilistic data structures (sketches) for Apache Flink that enables approximate, memory-efficient analytics over high-volume data streams — and am building ProjectASAP, low-latency data pipelines for agentic workloads.

Industry: I’ve interned at S&P Global on the Core Infra Team (observability, Kubernetes, service mesh), at ZKH Industrial Supply on ML Infrastructure, United Imaging Intelligence on backend systems, and built agentic systems using AutoGen at EnterviewAI.

Teaching: I’ve TA’d Distributed Systems (15-440/640) and Computer Systems (18-213/613) at CMU.