PeopleofSuperintelligence

Researcher

Richard Shin

Also known as Eui Chul Shin, 신의철

Richard Shin is a research scientist at Google DeepMind who works on neural program synthesis, semantic parsing, and coding models. His 2017 paper with Jonathon Cai and Dawn Song, 'Making Neural Programming Architectures Generalize via Recursion', received an ICLR 2017 Best Paper award. At UC Berkeley, he completed a PhD in computer science under Dawn Song in 2020; RAT-SQL, a text-to-SQL parser he coauthored, appeared at ACL 2020. He was first author of 'Constrained Language Models Yield Few-Shot Semantic Parsers' in 2021 and worked as a Principal Researcher at Microsoft Semantic Machines on conversational AI built with large language models. He now works on post-training for Gemini's coding capabilities, with a focus on SWE agents.

Badges

Timeline

  1. 2010–2015

    Security papers with Dawn Song

    His early research covers security applications of machine learning, software security, and web security. In 2010, he coauthors 'The Emperor's New APIs: On the (In)Secure Usage of New Client-side Primitives' with Dawn Song and others, and in 2015 he coauthors 'Recognizing Functions in Binaries with Neural Networks' with Dawn Song and Reza Moazzezi for USENIX Security.

  2. 2017

    Best paper at ICLR 2017

    With Jonathon Cai and Dawn Song, he writes 'Making Neural Programming Architectures Generalize via Recursion', which adds recursion to the Neural Programmer-Interpreter framework and tests it on grade-school addition, bubble sort, topological sort, and quicksort. The paper receives a Best Paper award at ICLR 2017.

  3. 2018–2019

    Program synthesis from traces and idioms

    In 2018, he coauthors 'Improving Neural Program Synthesis with Inferred Execution Traces' with Illia Polosukhin and Dawn Song, a spotlight presentation at NeurIPS. In 2019, he is first author of Patois at NeurIPS, a system that mines common code idioms from a corpus and trains a tree-based neural synthesizer to use them.

  4. July 2020

    The RAT-SQL parser

    In July 2020, he coauthors RAT-SQL with Bailin Wang, Xiaodong Liu, Oleksandr Polozov, and Matthew Richardson at ACL. The text-to-SQL parser uses relation-aware self-attention for schema encoding and linking, and reaches 57.2% exact match on the Spider dataset, or 65.6% when augmented with BERT.

  5. August 2020

    A Berkeley PhD in program synthesis

    He completes his PhD in computer science at UC Berkeley in August 2020, advised by Dawn Song, with the dissertation 'Abstraction, Generalization, and Embodiment in Neural Program Synthesis'. It covers execution traces for synthesizing Karel programs, synthetic training data, text-to-SQL semantic parsing, and learned code idioms. A member of the Berkeley AI Research Lab, he also holds MS and BS degrees from Berkeley.

  6. 2021

    Few-shot semantic parsers

    In 2021, he is first author of 'Constrained Language Models Yield Few-Shot Semantic Parsers' at EMNLP, with Dan Klein, Jason Eisner, Benjamin Van Durme, and others. The paper uses large pretrained language models to paraphrase inputs into a controlled sublanguage that maps automatically to a target meaning representation.

  7. As of October 2026

    Research scientist at DeepMind

    He is a Research Scientist at Google DeepMind, where he works on post-training for Gemini's coding capabilities with a focus on SWE agents; earlier at Google, he worked on Jules. Before Google, he was a Principal Researcher at Microsoft Semantic Machines, where his work used large language models to build conversational AI systems at scale. In July 2025, he is among the authors of the Gemini 2.5 report.

Career

  1. NowGoogle DeepMindResearch Scientist
  2. Semantic MachinesPrincipal Researcher

Education

  1. 2020University of California, BerkeleyBS, MS, and PhD

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Richard Shin. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Richard Shin. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Richard Shin yet. Send us a free-licensed photo.

  1. Richard Shin rshin.github.io. Checked 22 October 2026.
  2. Making Neural Programming Architectures Generalize via Recursion arxiv.org. Checked 22 October 2026.
  3. AI Best Paper Awards aibestpape.rs. Checked 22 October 2026.
  4. Program Synthesis and Semantic Parsing with Learned Code Idioms proceedings.neurips.cc. Checked 22 October 2026.
  5. RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers aclanthology.org. Checked 22 October 2026.
  6. Abstraction, Generalization, and Embodiment in Neural Program Synthesis www2.eecs.berkeley.edu. Checked 22 October 2026.
  7. Constrained Language Models Yield Few-Shot Semantic Parsers aclanthology.org. Checked 22 October 2026.
  8. Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities arxiv.org. Checked 22 October 2026.