PeopleofSuperintelligence

Researcher

Piero Kauffmann

Piero Kauffmann is a Microsoft Research engineer whose work spans small language models, neural architecture search and reinforcement learning. He coauthored phi-1 in June 2023 and contributed to Phi-2 in December 2023; he later coauthored technical reports on Phi-3, Phi-4 and Phi-4-reasoning. In May 2026, he coauthored ECHO, a training objective that helps terminal agents learn from their environment. As of October 2026, he is a Member of Technical Staff and research engineer with Microsoft Research's AI Frontiers Lab, focused on LLM reasoning, infrastructure and reinforcement learning.

Works
5
Citations
335
h-index
3

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. By 2023

    Creating and maintaining Archai

    By 2023, he helps create and maintain Archai, a Microsoft Research project for neural architecture search.

  2. June 2023

    Introducing phi-1

    In June 2023, he coauthors Textbooks Are All You Need, which introduces phi-1, a Transformer-based model for code with 1.3 billion parameters.

  3. December 2023

    Contributing to Phi-2

    In December 2023, he is among the contributors to Phi-2, a 2.7 billion-parameter language model that demonstrates reasoning and language-understanding capabilities.

  4. April 2024

    A Phi-3 technical report

    In April 2024, he coauthors the Phi-3 Technical Report, which introduces phi-3-mini, a 3.8 billion-parameter model trained on 3.3 trillion tokens whose performance rivals Mixtral 8x7B and GPT-3.5.

  5. December 2024

    Phi-4 and synthetic data

    In December 2024, he coauthors the Phi-4 Technical Report, which presents a 14-billion-parameter model built with a recipe focused on data quality and synthetic data throughout training.

  6. April 2025

    Phi-4 reasoning models

    In April 2025, he coauthors the Phi-4-reasoning Technical Report, which introduces a 14-billion-parameter reasoning model trained by supervised fine-tuning of Phi-4. The report also develops Phi-4-reasoning-plus, which adds a short phase of outcome-based reinforcement learning.

  7. May 2026

    ECHO for terminal agents

    In May 2026, he coauthors ECHO (Environment Cross-entropy Hybrid Objective), which adds an auxiliary loss for predicting environment observation tokens to the policy-gradient loss used to train command-line agents. On TerminalBench-2.0, ECHO doubles GRPO pass@1 for Qwen3-8B from 2.70% to 5.17%.

  8. As of October 2026

    Member of Technical Staff

    As of October 2026, he is a Member of Technical Staff and research engineer at Microsoft Research, working with its AI Frontiers Lab. He continues to contribute to the Phi series and focuses on reasoning in large language models, LLM infrastructure and reinforcement learning.

Career

  1. NowMicrosoft ResearchMember of Technical Staff and research engineer
  2. 2023–2026Microsoft

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

Most-cited works

  1. Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone2024 · 164 citationsWith Adil Salim, Ahmed Awadallah, Ce Liu, Corby Rosset, Gustavo de Rosa and Harkirat Behl
  2. Textbooks Are All You Need2023 · 138 citationsWith Adam Tauman Kalai, Adil Salim, Gustavo de Rosa, Harkirat Behl, Mojan Javaheripi and Olli Saarikivi
  3. Phi-4 Technical Report2024 · 31 citationsWith Adil Salim, Cyril Zhang, Dingli Yu, Eric Price, Gustavo de Rosa and Harkirat Behl
  4. Phi-4-reasoning Technical Report2025 · 2 citationsWith Ahmed Awadallah, Gustavo de Rosa, Harkirat Behl, Mojan Javaheripi, Neel Joshi and Olli Saarikivi
  5. ECHO: Terminal Agents Learn World Models for Free2026 · 0 citationsWith Ahmed Awadallah

From OpenAlex, as of 21 October 2026.

  1. Among employers with 30+ full-time cases, OpenAI leads primary technical role H-1B pay at $310,000, and Amazon, Google and Microsoft file 57% of all casesIncludes Microsoft. Across 440,669 certified US H-1B filings from 44 employers (FY2020–FY2026), OpenAI leads primary technical role base pay at $310,000 and xAI leads overall base pay at $290,000.
  2. 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 each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Microsoft. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. Google has the highest combined h-index among AI organizationsIncludes Microsoft. 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 with dated positions at the same organization in the same years.

Sources and credits

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

  1. Contact — Archai microsoft.github.io. Checked 21 October 2026.
  2. Piero Kauffmann at Microsoft Research microsoft.com. Checked 21 October 2026.
  3. [2306.11644] Textbooks Are All You Need arxiv.org. Checked 21 October 2026.
  4. Phi-2: The surprising power of small language models - Microsoft Research microsoft.com. Checked 21 October 2026.
  5. [2404.14219] Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone arxiv.org. Checked 21 October 2026.
  6. Phi-4 Technical Report - Microsoft Research microsoft.com. Checked 21 October 2026.
  7. [2504.21318] Phi-4-reasoning Technical Report arxiv.org. Checked 21 October 2026.
  8. [2605.24517] ECHO: Terminal Agents Learn World Models for Free arxiv.org. Checked 21 October 2026.