Summary
Ansong Ni is a research scientist who teaches large language models to reason in both natural and formal languages, specializing in language-to-code and programmatic problem solving. With a PhD from Yale and internships at DeepMind, Meta AI, Microsoft, and AI2, he builds training and evaluation pipelines, curates data, and designs learning algorithms that improve LLM reasoning and execution. His work—published at ICLR, ICML, ACL and EMNLP—bridges academic rigor and industry-scale experimentation across FAIR and DeepMind teams. Based in New Haven, he brings 11 years of experience combining empirical analysis with practical tooling for model evaluation, and often explores partial-correctness and self-sampling techniques to boost math and code reasoning.
11 years of coding experience
Computer Science, Computer Science at University of California, Berkeley
Bachelor's degree, Computer Software Engineering, Bachelor's degree, Computer Software Engineering at Nanjing University
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Yale University
Master of Science in Computer Science, Computer Science, Master of Science in Computer Science, Computer Science at Carnegie Mellon University
Chinese, English