PeopleofSuperintelligence

Researcher

Mariana Raykova

Also known as Mariana Petrova Raykova

Mariana Raykova is a cryptographer whose work applies privacy protections to machine learning and language models. Before Google, she earned a computer-science PhD at Columbia, completed a postdoctoral year in IBM Research Watson's Cryptography Group, worked as a research scientist at SRI International and served as an assistant professor at Yale. She coauthored a 2024 watermarking scheme for open-source language models. Her 2025 work includes secure aggregation for federated learning, defenses for local differential privacy and a Google patent for cryptographic multi-party reach and frequency computation. She is a research scientist at Google and coauthored Hadal in 2026, a framework for encrypted computation and model training with label differential privacy.

Now at
Google
Badges

Timeline

  1. By 2013

    A Columbia PhD and IBM postdoc

    Before joining SRI in 2013, she earns a computer-science PhD at Columbia and spends a year as a postdoc in IBM Research Watson's Cryptography Group.

  2. 2013–2015

    Research scientist at SRI

    At SRI International's Computer Science Laboratory, she works as a research scientist from 2013 to 2015.

  3. 2016–2018

    Assistant professor at Yale

    She serves as an assistant professor in Yale's Department of Computer Science from 2016 to 2018. In 2017, her group develops cryptographic techniques to let parties share selected private data, and she receives a Google Faculty Research Award for the project.

  4. 2019

    Joining Google Research

    She joins Google as a research scientist in 2019. Her research includes secure computation, oblivious data structures, zero knowledge and verifiable computation.

  5. 2024

    Watermarks for open language models

    In 2024, she coauthors a watermarking scheme for open-source language models that modifies model parameters and lets detectors identify generated text. The authors evaluate it on OPT-1.3B and OPT-6.7B.

  6. 2025

    Privacy tools for machine learning

    At the 2025 SysteMPC workshop, she presents Willow, a secure aggregation framework for single-shot clients, drawing on private analytics for federated learning and exposure notifications. In 2025, she coauthors Vεrity, which presents two methods to prevent poisoning attacks against local differential privacy while preserving its guarantees. A Google patent granted in 2025 lists her as a co-inventor on cryptographic computation for multi-party reach and frequency.

  7. 2026

    Research scientist at Google

    She is a research scientist at Google. In 2026, she coauthors Hadal, an encrypted-computation framework, and Hadal-ml, a machine-learning package for model training with label differential privacy. The work integrates with TensorFlow and avoids a trusted third party.

Career

  1. From 2019GoogleResearch scientist
  2. 2016–2018Yale UniversityAssistant professor
  3. 2013–2015SRI InternationalResearch scientist
  4. IBMPostdoctoral researcher

Education

  1. Columbia UniversityComputer science PhD

From the biography.

  1. 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.
  2. 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.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Google. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

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

Sources and credits

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

  1. Mariana Raykova research.google. Checked 21 October 2026.
  2. Engineering faculty devising solutions to increasing cybersecurity threats | Yale News news.yale.edu. Checked 21 October 2026.
  3. Provably Robust Watermarks for Open-Source Language Models arxiv.org. Checked 21 October 2026.
  4. SysteMPC Workshop 2025 bu.edu. Checked 21 October 2026.
  5. V𝜖rity: Verifiable Local Differential Privacy research.google. Checked 21 October 2026.
  6. US12395317B2 - Cryptographic computation techniques for multi-party reach and frequency - Google Patents patents.google.com. Checked 21 October 2026.
  7. Hadal: Centralized Label DP Training without a Trusted Party research.google. Checked 21 October 2026.