PeopleofSuperintelligence

Researcher

Andres Codas

Andres Codas is a Microsoft research software engineer whose publications span neural simulation, code generation, and language-model training. He earned an Engineering Cybernetics PhD at the Norwegian University of Science and Technology in 2016 and coauthored a 2019 IBM Research paper on neural proxies for oil-reservoir simulation. His 2022 CodeRanker paper addressed code correctness; Orca 2 in 2023 explored small-model reasoning, and AgentInstruct in 2024 introduced synthetic training data. As of October 2026, he is a Senior Research Software Engineer on the Research at Redmond Data Science team at Microsoft Research.

Works
38
Citations
306
h-index
10
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2016

    An Engineering Cybernetics PhD

    He earns a PhD in Engineering Cybernetics at the Norwegian University of Science and Technology in 2016. During his education, he conducts applied research in control engineering and optimization in the physics-AI domain.

  2. 2019

    Neural proxies for oil reservoirs

    In 2019, he coauthors a neural-network proxy study of oil-reservoir simulation at IBM Research. The paper reports a practical speedup of more than 2,000 times over the simulator.

  3. Since 2021

    Microsoft language-model workflows

    He joins Microsoft in 2021 and develops workflows for collecting and wrangling data to train and evaluate large language models.

  4. 2022

    Fault-aware neural code rankers

    In 2022, he coauthors Fault-Aware Neural Code Rankers, which introduces CodeRanker, a neural ranker that predicts sampled program correctness without executing it.

  5. 2023

    Orca 2 reasoning models

    In 2023, he coauthors Orca-2: Teaching Small Language Models How to Reason, a Microsoft Research paper about small-model reasoning.

  6. 2024

    AgentInstruct training data

    In 2024, he coauthors AgentInstruct, an agentic framework for creating synthetic training data. The work creates a 25-million-pair dataset to teach language models skills such as tool use, coding, and reading comprehension.

  7. As of October 2026

    Senior Research Software Engineer

    As of October 2026, he is a Senior Research Software Engineer on Microsoft's Research at Redmond Data Science team.

Career

  1. NowMicrosoft ResearchSenior Research Software Engineer
  2. 2022–2024Microsoft
  3. 2017–2021IBM
  4. 2017–2021IBM Research - Thomas J. Watson Research Center
  5. 2008–2012Universidade Federal de Santa Catarina

Education

  1. 2016Norwegian University of Science and TechnologyPhD

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Integrated production optimization of oil fields with pressure and routing constraints: The Urucu field2012 · Computers & Chemical Engineering · 50 citations
  2. Mixed-integer linear optimization for optimal lift-gas allocation with well-separator routing2011 · European Journal of Operational Research · 44 citations
  3. Orca 2: Teaching Small Language Models How to Reason2023 · 30 citationsWith Ahmed Awadallah and Anastasia Razdaibiedina
  4. Fault-Aware Neural Code Rankers2022 · neural information processing systems · 22 citationsWith Jianfeng Gao
  5. Closed-Loop Predictions in Reservoir Management Under Uncertainty2017 · SPE Journal · 21 citations

From OpenAlex, as of 18 October 2026.

  1. Google alumni founded or led 82 other organizations, the most among the labs we compareIncludes Andres Codas. We trace 189 organizations founded or led by AI lab alumni, and 196 career moves between the labs.
  2. The company share of AI researchers’ top papers rose from 3% to 32%Includes Microsoft. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 3% in 1975–1979 to 32% in 2020–2024.
  3. Google has the highest combined h-index among AI organizationsIncludes Microsoft. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  4. Google had the most people shaping AI every year from 2014 to 2025Includes Microsoft. Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Andres Codas at Microsoft Research microsoft.com. Checked 18 October 2026.
  2. Accelerating Physics-Based Simulations Using End-to-End Neural Network Proxies: An Application in Oil Reservoir Modeling frontiersin.org. Checked 18 October 2026.
  3. Fault-Aware Neural Code Rankers proceedings.neurips.cc. Checked 18 October 2026.
  4. Orca - Microsoft Research: Publications microsoft.com. Checked 18 October 2026.
  5. AgentInstruct: Toward Generative Teaching with Agentic Flows arxiv.org. Checked 18 October 2026.