PeopleofSuperintelligence

Builder

Thomas T. Joshi

Thomas T. Joshi is an AI builder whose work spans language-model programming, inference systems and coding-agent evaluation. In 2023, he coauthored DSPy, a programming model that represents language-model pipelines as text-transformation graphs, and in February 2024 he hosted a podcast episode about AI in video understanding. His 2025 papers address long-context inference and continual learning for coding agents, after his first job building conversational-AI infrastructure for healthcare. He is an Associate on NEA's Technology Investing Team, which he joined in 2025, and coauthored two 2026 NEA articles on research-first AI and CuspAI's materials work.

Now at
New Enterprise Associates
Links
LinkedIn

Timeline

  1. 2023

    DSPy for language-model pipelines

    In 2023, he coauthors DSPy, which models language-model pipelines as text-transformation graphs. Its compiler optimizes a pipeline to maximize a chosen metric.

  2. February 2024

    Hosting an AI podcast

    In February 2024, he hosts an episode of the Collective Intelligence Community Podcast with Stephen Campbell. They interview Twelve Labs CTO Aidan Lee about AI in video understanding.

  3. By 2025

    AI engineering for healthcare

    By 2025, his first job is as an AI Engineer building infrastructure to support language and classification models for conversational AI in healthcare.

  4. June 2025

    PagedAttention meets FlexAttention

    In June 2025, he coauthors a paper that integrates PagedAttention with PyTorch's FlexAttention in IBM's Foundation Model Stack. Its benchmarks report latency growth of about two times from 128 to 2,048 tokens with a global key-value cache.

  5. June 2025

    A benchmark for coding agents

    In June 2025, he coauthors SWE-Bench-CL with Shayan Chowdhury and Fatih Uysal. It arranges GitHub issues in chronological sequences to evaluate how coding agents accumulate experience, transfer knowledge and resist forgetting.

  6. June 2026

    The Neolab Wild West

    In June 2026, he coauthors “The Neolab Wild West” with Madison Faulkner, Lila Tretikov and Andrew Schoen. The essay argues that specialized, research-first AI companies can achieve venture-like outcomes.

  7. July 2026

    A CuspAI investment update

    In July 2026, he coauthors NEA's “From Algorithms to Atoms, Part II: Doubling Down on CuspAI,” an update on the company's AI materials work.

  8. Since 2025

    Associate at NEA

    In 2025, he joins NEA as an Associate on its Technology Investing Team, focused on early- and growth-stage investments. He holds a B.Sc. in Artificial Intelligence from Columbia University.

Career

  1. Since 2025New Enterprise AssociatesAssociate

Education

  1. Columbia UniversityB.Sc.

From the biography.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Columbia University. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. Nonprofit roles are 2.3 times as common in AI safetyIncludes Columbia University. We estimate where people focused on safety, alignment and evaluation work. Nonprofit roles appear for 25.1% of safety people with known affiliations, compared with 11.0% in other areas.
  3. Universities had a net outflow of 2,068 moves in AI careersIncludes Columbia University. Our estimates of career flows and median h-index across universities and companies in AI, including the role of joint appointments and earlier academic jobs.
  4. People shaping AI work at a median of 3 organizations over a career: 3 for university starters and institute and lab starters, versus 2 for company startersIncludes Columbia University. Across 6,853 people shaping AI with a dated work role, we estimate a median of 3 parent organizations over a career (mean 3.6), comparing role groups, starting sectors and starting decades from the 1950s to 2020–2026.

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 Thomas T. Joshi yet. Send us a free-licensed photo.

  1. DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines arxiv.org. Checked 21 October 2026.
  2. DSPy: Compiling Declarative Language Model Calls into State-of-the-Art Pipelines proceedings.iclr.cc. Checked 21 October 2026.
  3. Twelve Labs, CTO - Aidan Lee | Ep 10 podcasts.apple.com. Checked 21 October 2026.
  4. Thomas Joshi nea.com. Checked 21 October 2026.
  5. Paged Attention Meets FlexAttention: Unlocking Long-Context Efficiency in Deployed Inference arxiv.org. Checked 21 October 2026.
  6. SWE-Bench-CL: Continual Learning for Coding Agents arxiv.org. Checked 21 October 2026.
  7. The Neolab Wild West nea.com. Checked 21 October 2026.
  8. From Algorithms to Atoms, Part II: Doubling Down on CuspAI nea.com. Checked 21 October 2026.