---
id: "4d0171dd-4ad7-4dd6-89de-4161834457e3"
type: "Person"
name: "Diederik P. Kingma"
slug: "diederik-p-kingma"
canonical: "https://groundingwiki.com/en/people/diederik-p-kingma/"
markdown: "https://groundingwiki.com/en/people/diederik-p-kingma/index.md"
language: en
last_reviewed: 2026-09-28
same_as:
  - "https://wikidata.org/wiki/Q29022624"
  - "https://wikidata.org/wiki/Q21708200"
---

# Diederik P. Kingma

Diederik P. “Durk” Kingma is a computer scientist and machine-learning researcher whose work includes variational inference, generative modeling, and stochastic optimization; he was a member of OpenAI’s founding research team. [1]

## Core facts

- **Entity type:** Person

- **Employer:** [OpenAI](https://groundingwiki.com/en/organizations/openai/index.md) — [1] first-party, checked 30 Sept 2026

- **Date of birth:** 1983 — [2] first-party, checked 28 Sept 2026

- **Occupations:** Computer scientist; machine-learning researcher — [1] first-party, checked 30 Sept 2026

- **Education:** Earned a PhD in computer science at the University of Amsterdam in 2017, with distinction \(cum laude\); his thesis was advised by Max Welling. — [2] first-party, checked 28 Sept 2026

- **Official profile or website:** [dpkingma.com](https://dpkingma.com/) — [1] first-party, checked 30 Sept 2026

## Who is Diederik P. Kingma?

Kingma’s research has focused on scalable machine learning, variational inference, and generative modeling. His co-authored work includes Auto-Encoding Variational Bayes, which helped establish the variational autoencoder approach; the Adam stochastic optimization method; the Glow generative flow model; and Variational Diffusion Models. He earned a computer-science PhD with distinction at the University of Amsterdam in 2017. Professionally, he was part of OpenAI’s founding research team, later worked at Google Brain / DeepMind, and his biography records research at Anthropic from 2024 onward. — [2] first-party, checked 28 Sept 2026; [4] structured public data, checked 28 Sept 2026; [5] structured public data, checked 28 Sept 2026; [6] first-party, checked 3 Oct 2026; [1] first-party, checked 30 Sept 2026

## History

- **2015:** Was part of OpenAI’s founding team and worked there as a research scientist from 2015 to 2018. — [1] first-party, checked 30 Sept 2026

- **2018:** Worked as a research scientist at Google Brain / DeepMind from 2018 to 2024. — [1] first-party, checked 30 Sept 2026

- **2024:** Since 2024, has conducted research on large-scale machine learning at Anthropic. — [1] first-party, checked 30 Sept 2026

## Education

- **Phd:** Earned a PhD in computer science at the University of Amsterdam in 2017, with distinction \(cum laude\); his thesis was advised by Max Welling. — [2] first-party, checked 28 Sept 2026

- **Phd Thesis:** His doctoral thesis, “Variational Inference and Deep Learning: A New Synthesis,” was defended at the University of Amsterdam in 2017. — [7] first-party, checked 28 Sept 2026

## Career and roles

- **Advanza:** Co-founded Advanza and served as its technical lead from 2010 to 2012; his biography describes a successful exit in 2016. — [1] first-party, checked 30 Sept 2026

- **Anthropic:** Has conducted research on large-scale machine learning at Anthropic since 2024. — [1] first-party, checked 30 Sept 2026

- **Google Brain Deepmind:** Worked as a research scientist at Google Brain / DeepMind from 2018 to 2024, leading research projects on generative models for text, images, and video. — [1] first-party, checked 30 Sept 2026

- **New York University:** Worked as a junior research scientist in Yann LeCun’s lab at New York University in 2009 and 2012. — [1] first-party, checked 30 Sept 2026

- **OpenAI:** Was part of OpenAI’s founding research team and worked as a research scientist there from 2015 to 2018; he led its Algorithms team. — [1] first-party, checked 30 Sept 2026

## Contributions and work

- **Adam Optimizer:** Co-authored “Adam: A Method for Stochastic Optimization” with Jimmy Ba. — [5] structured public data, checked 28 Sept 2026

- **Adam Optimizer:** Co-authored “Adam: A Method for Stochastic Optimization” with Jimmy Ba; the paper introduced an adaptive first-order optimization algorithm for stochastic objectives. — [5] structured public data, checked 28 Sept 2026

- **Glow:** Co-authored Glow, a generative-flow model using invertible 1×1 convolutions for image synthesis and exact likelihood modeling. — [8] structured public data, checked 28 Sept 2026

- **Inverse Autoregressive Flow:** Co-authored “Improved Variational Inference with Inverse Autoregressive Flow,” a method for more flexible approximate posterior distributions. — [9] structured public data, checked 28 Sept 2026

- **Variational Autoencoder:** Co-authored “Auto-Encoding Variational Bayes” with Max Welling, a paper that introduced a scalable variational inference and learning approach foundational to variational autoencoders. — [4] structured public data, checked 28 Sept 2026

- **Variational Autoencoder:** Co-authored “Auto-Encoding Variational Bayes,” a paper introducing a scalable variational-inference and learning method that helped establish the variational autoencoder \(VAE\). — [4] structured public data, checked 28 Sept 2026

- **Variational Diffusion Models:** Co-authored research introducing Variational Diffusion Models, including a method for learning the diffusion noise schedule jointly with model parameters. — [10] structured public data, checked 28 Sept 2026

- **Focus:** His research focuses on scalable machine-learning methods, variational inference, and generative modeling. — [1] first-party, checked 30 Sept 2026

- **Focus:** Scalable machine learning and generative modelling. — [1] first-party, checked 30 Sept 2026

## Awards and recognition

- **Dutch Data Science 2019:** Received the 2019 Gerrit van Dijk Prize \(Dutch Data Science Award\) for his dissertation research. — [11] first-party, checked 28 Sept 2026

- **Ellis Phd 2019:** Received one of the inaugural ELLIS PhD Awards in 2019 for dissertation-phase research in machine learning and artificial intelligence. — [12] first-party, checked 28 Sept 2026

- **Google Doctoral Fellowship 2015:** Received Google’s first European Doctoral Fellowship in Deep Learning in 2015. — [1] first-party, checked 30 Sept 2026

- **Iclr 2024:** The 2024 ICLR Test of Time Award recognized “Auto-Encoding Variational Bayes,” co-authored with Max Welling. — [13] first-party, checked 28 Sept 2026

- **Iclr 2025:** The 2025 ICLR Test of Time Award recognized “Adam: A Method for Stochastic Optimization,” co-authored with Jimmy Ba. — [14] first-party, checked 28 Sept 2026

## Relationships

- **created:** [OpenAI](https://groundingwiki.com/en/organizations/openai/index.md) — [15] structured public data, checked 28 Sept 2026

- **founded:** [Advanza B.V.](https://groundingwiki.com/en/companies/advanza/index.md) — [1] first-party, checked 30 Sept 2026

- **worked at:** [OpenAI](https://groundingwiki.com/en/organizations/openai/index.md) — [1] first-party, checked 30 Sept 2026

- **worked at:** [Google DeepMind](https://groundingwiki.com/en/organizations/google-deepmind/index.md) — [1] first-party, checked 30 Sept 2026

- **worked at:** [Advanza B.V.](https://groundingwiki.com/en/companies/advanza/index.md) — [1] first-party, checked 30 Sept 2026

- **founded:** [OpenAI](https://groundingwiki.com/en/organizations/openai/index.md) — [6] first-party, checked 3 Oct 2026

- **works at:** [Anthropic](https://groundingwiki.com/en/companies/anthropic/index.md) — [16] independent, checked 28 Sept 2026

## Sources
1. [Diederik P. \(Durk\) Kingma — Brief Bio](https://dpkingma.com/)
   - Publisher: Diederik P. Kingma
   - Type: first-party
   - Checked: 30 Sept 2026
   - Evidence:
     - Definition: "Kingma describes himself as a computer scientist and researcher; OpenAI’s 2015 founding announcement lists Durk Kingma among its founding research engineers and scientists."
     - Employer: "Kingma’s career timeline records his OpenAI research-scientist role from 2015 to 2018."
     - Occupations: "Kingma’s official biography identifies him as a computer scientist and researcher focused on scalable machine learning and generative modeling."
     - Official profile or website: "Kingma’s official biography links to dpkingma.com as his personal website."
     - Who is Diederik P. Kingma?: "Kingma’s biography documents his research areas, named contributions, education, and career chronology; the cited papers and institutional sources support the individual research and education details."
     - 2015: "Kingma’s official employment timeline lists his OpenAI founding-team and research-scientist period as 2015–2018."
     - 2018: "Kingma’s official employment timeline records a research-scientist role at Google Brain / DeepMind from 2018 through 2024."
     - 2024: "Kingma’s current official biography lists research on large-scale machine learning at Anthropic from 2024 onward."
     - Advanza: "Kingma’s official biography records his Advanza co-founder and technical-lead role for 2010–2012 and describes a later successful exit without specifying an acquisition."
     - Anthropic: "Kingma’s current official biography lists research on large-scale machine learning at Anthropic from 2024 onward."
     - Google Brain Deepmind: "Kingma’s official career timeline dates this research-scientist role to 2018–2024 and describes the generative-model research areas."
     - New York University: "Kingma’s official career timeline lists junior research-scientist work in Yann LeCun’s lab at NYU in 2009 and 2012."
     - OpenAI: "Kingma’s official career timeline records his founding-team role, 2015–2018 research-scientist tenure, and leadership of the Algorithms team."
     - Focus: "Kingma’s official biography states that his research focuses on scalable methods for machine learning and generative modeling."
     - Focus: "The official biography describes his research focus as scalable machine learning and generative modeling."
     - Google Doctoral Fellowship 2015: "Kingma’s official biography lists the 2015 European Doctoral Fellowship in Deep Learning as an award and honor."
     - founded: "Kingma’s official biography identifies him as an Advanza co-founder and dates his role to 2010–2012."
     - worked at: "Kingma’s career timeline records his OpenAI research-scientist role from 2015 to 2018."
     - worked at: "Kingma’s career timeline records his Google Brain / DeepMind research-scientist role from 2018 to 2024."
     - worked at: "Kingma’s official biography lists his Advanza co-founder and technical-lead role from 2010 to 2012."
2. [Album Academicum: D.P. Kingma](https://albumacademicum.uva.nl/en/id/id170704)
   - Publisher: University of Amsterdam
   - Type: first-party
   - Checked: 28 Sept 2026
   - Evidence:
     - Date of birth: "The University of Amsterdam Album Academicum lists D.P. Kingma’s birth year as 1983."
     - Education: "The University of Amsterdam academic record lists a doctoral degree with distinction in 2017, the thesis title, and Max Welling as supervisor."
     - Who is Diederik P. Kingma?: "The University record documents Kingma’s doctoral education and distinction."
     - Phd: "The University of Amsterdam academic record lists a doctoral degree with distinction in 2017, the thesis title, and Max Welling as supervisor."
3. [Diederik P. Kingma — Wikidata](https://wikidata.org/wiki/Q29022624)
   - Publisher: Wikidata
   - Type: structured public data
   - Checked: 22 Sept 2026
4. [Auto-Encoding Variational Bayes](https://arxiv.org/abs/1312.6114)
   - Publisher: arXiv
   - Type: structured public data
   - Checked: 28 Sept 2026
   - Evidence:
     - Who is Diederik P. Kingma?: "The original VAE paper and Kingma’s biography document his generative-model research; UvA, OpenAI, ICLR, and ELLIS sources substantiate the education, career, and awards summarized here."
     - Variational Autoencoder: "The arXiv record lists Kingma and Welling as authors and describes their variational inference and learning method; ICLR later recognized the paper with its 2024 Test of Time award."
     - Variational Autoencoder: "The arXiv record lists Kingma and Max Welling as authors and describes the paper’s stochastic variational-inference and learning contribution."
5. [Adam: A Method for Stochastic Optimization](https://arxiv.org/abs/1412.6980)
   - Publisher: arXiv
   - Type: structured public data
   - Checked: 28 Sept 2026
   - Evidence:
     - Who is Diederik P. Kingma?: "The original Adam paper documents the optimization contribution summarized in the overview."
     - Adam Optimizer: "The arXiv record lists Diederik P. Kingma and Jimmy Ba as authors of the paper introducing the Adam stochastic optimization method."
     - Adam Optimizer: "The arXiv paper lists Kingma and Jimmy Ba as authors and states that it introduces Adam for first-order stochastic optimization."
6. [Introducing OpenAI](https://openai.com/index/introducing-openai)
   - Publisher: OpenAI
   - Type: first-party
   - Checked: 3 Oct 2026
   - Evidence:
     - Who is Diederik P. Kingma?: "OpenAI’s founding announcement lists Kingma among the founding research engineers and scientists."
     - founded: "OpenAI’s 2015 launch announcement explicitly lists Durk Kingma among the organization’s founding research engineers and scientists."
7. [Variational Inference and Deep Learning: A New Synthesis](https://pure.uva.nl/ws/files/17891313/Thesis.pdf)
   - Publisher: University of Amsterdam
   - Type: first-party
   - Checked: 28 Sept 2026
   - Evidence:
     - Phd Thesis: "The deposited thesis identifies Kingma, gives the thesis title, and records its defense at the University of Amsterdam on 25 October 2017."
8. [Glow: Generative Flow with Invertible 1x1 Convolutions](https://arxiv.org/abs/1807.03039)
   - Publisher: arXiv
   - Type: structured public data
   - Checked: 28 Sept 2026
   - Evidence:
     - Glow: "The arXiv record lists Kingma and Prafulla Dhariwal as authors and describes Glow’s invertible 1×1 convolution approach."
9. [Improved Variational Inference with Inverse Autoregressive Flow](https://arxiv.org/abs/1606.04934)
   - Publisher: arXiv
   - Type: structured public data
   - Checked: 28 Sept 2026
   - Evidence:
     - Inverse Autoregressive Flow: "The arXiv record lists Kingma among the authors and presents inverse autoregressive flow for improving variational inference."
10. [Variational Diffusion Models](https://arxiv.org/abs/2107.00630)
   - Publisher: arXiv
   - Type: structured public data
   - Checked: 28 Sept 2026
   - Evidence:
     - Variational Diffusion Models: "The paper lists Kingma among its authors and describes Variational Diffusion Models and joint noise-schedule optimization."
11. [Jury report: Gerrit van Dijk Prize 2019](https://khmw.nl/wp-content/uploads/Juryrapport-Gerrit-van-Dijk-Prijs-2019.pdf)
   - Publisher: Royal Holland Society of Sciences and Humanities
   - Type: first-party
   - Checked: 28 Sept 2026
   - Evidence:
     - Dutch Data Science 2019: "The Royal Holland Society of Sciences and Humanities’ jury report identifies Diederik Kingma as the 2019 Gerrit van Dijk Prize recipient for his dissertation."
12. [ELLIS Announces First Six Winners of the ELLIS PhD Award](https://ellis.eu/news/ellis-phd-award)
   - Publisher: European Laboratory for Learning and Intelligent Systems
   - Type: first-party
   - Checked: 28 Sept 2026
   - Evidence:
     - Ellis Phd 2019: "ELLIS’s 2019 announcement names Diederik Kingma of the University of Amsterdam among its first six PhD Award winners."
13. [ICLR 2024 Test of Time Award](https://blog.iclr.cc/2024/05/07/iclr-2024-test-of-time-award)
   - Publisher: International Conference on Learning Representations
   - Type: first-party
   - Checked: 28 Sept 2026
   - Evidence:
     - Iclr 2024: "ICLR’s official award announcement names Kingma and Welling as authors of the 2024 Test of Time award-winning paper."
14. [Announcing the Test of Time Award Winners from ICLR 2015](https://blog.iclr.cc/2025/04/14/announcing-the-test-of-time-award-winners-from-iclr-2015)
   - Publisher: International Conference on Learning Representations
   - Type: first-party
   - Checked: 28 Sept 2026
   - Evidence:
     - Iclr 2025: "ICLR’s official 2025 award announcement names Kingma and Jimmy Ba as authors of the Test of Time winning Adam paper."
15. [OpenAI \(Q21708200\)](https://wikidata.org/wiki/Q21708200)
   - Publisher: Wikidata
   - Type: structured public data
   - Checked: 28 Sept 2026
   - Evidence:
     - created: "Wikidata P112 links OpenAI to Diederik P. Kingma."
16. [Anthropic hires OpenAI co-founder Durk Kingma](https://techcrunch.com/2024/10/01/anthropic-hires-openai-co-founder-durk-kingma)
   - Publisher: TechCrunch
   - Type: independent
   - Checked: 28 Sept 2026
   - Evidence:
     - works at: "TechCrunch reported that Anthropic hired Kingma in October 2024; his official biography confirms research on large-scale machine learning there from 2024 onward."

HTML version: https://groundingwiki.com/en/people/diederik-p-kingma/
Methodology: https://groundingwiki.com/methodology/
Source policy: https://groundingwiki.com/source-policy/
Corrections: https://groundingwiki.com/corrections/?entity=Diederik%20P.%20Kingma
