Summary
Chiao-lun Cheng is a Machine Learning Engineer with 12 years of experience building production-grade ML systems, currently applying ML to drive superhuman selling at Nooks. He combines deep learning (Transformers, GNNs, LSTMs) and computer vision with robust engineering: CUDA streams, Kubernetes autoscaling-from-zero, Istio, and NATS.IO service buses for low-latency pipelines. At Scale AI he shipped end-to-end solutions from ETL and Athena analytics to form parsing with LayoutLM, metric learning, and LoRA-finetuned LLMs, and optimized opportunistic batching and dataloading for real-world throughput. His background spans algorithmic trading, causal inference and multilevel modeling, plus hands-on systems work (Numba, Cython, Pyspark, Mongo/Postgres) informed by a PhD in computational physics from MIT. Comfortable moving between research and production, he has a history of squeezing performance out of infrastructure and models while modeling labeler behavior and task difficulty with principled statistical methods. A less obvious strength is his fluency across paradigms—from parser combinators and Clojure servers to modern neural retrieval and active learning—making him effective at bridging legacy systems and cutting-edge ML.
12 years of coding experience
17 years of employment as a software developer
B.S. Chemistry Physics Biochemistry, B.S. Chemistry Physics Biochemistry at University of California, Berkeley
PhD Physical Chemistry Computational Physics, PhD Physical Chemistry Computational Physics at Massachusetts Institute of Technology
Dunman High
University of California, Irvine
English, Chinese