Ahsen Khaliq

wiki
, , , ,
1

Machine learning engineer and ML Growth Lead at Hugging Face, widely known by the handle AK (@_akhaliq), who curates daily AI research papers for one of the largest audiences in the field and builds open-source tooling in the Hugging Face ecosystem.

Profile

Field Detail
Full name Ahsen Khaliq
Known as AK, @_akhaliq
Date of birth Not publicly available
Nationality American
Location Washington DC–Baltimore Area
Institution Hugging Face
Role ML Growth Lead
Research interests AI/ML paper curation, open-source developer tooling, generative AI demos
Education University of Maryland; Georgia Institute of Technology (degree types and years not publicly confirmed)
X / Twitter @_akhaliq
GitHub AK391
Hugging Face akhaliq
Substack akhaliq.substack.com

Overview

Ahsen Khaliq, universally referred to as “AK” in AI communities, is a machine learning engineer and ML Growth Lead at Hugging Face whose decade-long practice of curating and sharing arXiv preprints on X (formerly Twitter) has made him one of the most consequential information gatekeepers in applied AI research. His account (@_akhaliq, joined April 2014) has accumulated nearly 500,000 followers, and a 2024 study by researchers at the University of California, Santa Barbara found that papers he shares receive median citation counts two to three times higher than a control group — a finding reported by IEEE Spectrum as evidence of a structural shift in how the research community discovers and values work. Beyond curation, Khaliq has built a suite of widely used open-source tools on Hugging Face Spaces, most notably AnyChat and AnyCoder, and is the individual behind Hugging Face’s official Daily Papers feed. He is based in the Washington DC–Baltimore area.

Early Life & Education

Khaliq attended the University of Maryland and later the Georgia Institute of Technology; specific degree titles and graduation years are not publicly confirmed. His early career, spanning roughly 2015 to 2019, included roles across sectors — software development at Cybrary, financial analysis at Wells Fargo, an industrial engineering internship at UPS, and computer vision research engineering at Uraniom — before he pivoted fully toward machine learning in 2020.

Career

Cybrary (2015–2016)

Software Developer. Early role at a cybersecurity training platform in the Washington DC area.

Wells Fargo (2016–2018)

Analyst (2016–2017), then Analyst II (2017–2018). Data and analytical work within a large financial institution.

UPS (2017)

Industrial Engineering Intern. A brief internship concurrent with or immediately following his analyst role.

Uraniom (May 2019 – October 2019)

Computer Vision Research Engineer. First dedicated ML role; Uraniom is a French startup focused on 3D avatar technology. This marks Khaliq’s transition into applied deep learning.

Snap Inc. (July 2020 – August 2020)

Machine Learning Residency. A short-term ML residency at Snap, providing exposure to production-scale computer vision and ML systems.

Runway (December 2020 – January 2021)

Software Developer. Brief engagement at Runway, the generative AI video company, at an early stage before the company’s public prominence.

Gradio (March 2021 – November 2021)

Machine Learning Engineer. Joined Gradio — the Python library for building and sharing ML demos — and contributed to its ML tooling. Gradio was subsequently acquired by Hugging Face, transitioning Khaliq into the acquiring organization.

Hugging Face (November 2021 – present)

Machine Learning Engineer (November 2021 – July 2023), then ML Growth Lead (July 2023 – present). Khaliq joined via the Gradio acquisition and has since become one of the most publicly visible members of the organization. In May 2023, after accumulating approximately 17,000 paper-sharing posts on X, he migrated his daily curation practice to a dedicated Hugging Face feature — the Daily Papers feed (huggingface.co/papers), branded “Your daily dose of AI research from AK.” He has shipped over 650 Hugging Face Spaces, 46 models, and 3 datasets, and opened a community paper submission pipeline (huggingface.co/papers/submit) that allows authors to self-submit preprints directly to the feed. As ML Growth Lead his responsibilities encompass platform growth, developer community engagement, and open-source tooling.

Key Contributions

  • Hugging Face Daily Papers — The official daily research digest at huggingface.co/papers, initiated and sustained by Khaliq. Papers surfaced here are indexed on the platform and linkable from model, dataset, and Space cards, embedding curation directly into the ML development workflow.

  • X paper curation (@_akhaliq) — Beginning around 2018, Khaliq posted approximately 17,000 tweets highlighting arXiv preprints across subfields including multimodal LLMs, video generation, reinforcement learning, and diffusion models. A UCSB study (arXiv:2401.13782, 2024) identified his account as one of two whose shares correlate with a two-to-three-fold increase in median citation counts relative to unshared papers.

  • AnyChat (GitHub: AK391/anychat) — An open-source unified chat interface built on Gradio aggregating access to dozens of frontier models — reasoning models, vision-language models, and open-weight releases — behind a single UI. Used by the community as a standard first-access point for newly released models.

  • AnyCoder (akhaliq/anycoder) — An open-source vibe-coding tool that generates working frontend code (HTML, React, Gradio, Streamlit, and other frameworks) from natural-language prompts, with multimodal generation from UI screenshots. Reported by VentureBeat as one of the first open-source alternatives to proprietary tools such as Lovable, and among the first to support Moonshot’s Kimi K2 model at launch.

  • ai-gradio (GitHub: AK391/ai-gradio) — A Python package that simplifies building AI-powered applications backed by multiple model providers through a unified Gradio interface; approximately 1,600 GitHub stars as of 2025–2026.

  • ML conference indexing on Hugging Face — Khaliq organized dedicated Hugging Face organizations for a series of major ML venues including ICML 2022, ECCV 2022, NAACL 2022, ICCV 2023, ICML 2023, ICLR 2024, SIGGRAPH 2022, and others, centralizing paper and demo discovery around the conference calendar.

Awards & Recognition

  • IEEE Spectrum feature (March 2024) — Profiled as one of two AI influencers with measurable impact on research citation patterns, in a piece by Julianne Pepitone titled “A Few Social Media Influencers Are Shaping AI.”

  • UCSB academic study (arXiv:2401.13782, January 2024) — Named as a primary case study in research examining social media’s structural influence on ML research visibility, with the authors describing him and one peer as figures “akin to journalists in civic society, highlighting and contextualizing significant works for the community.”

  • Hugging Face platform recognition — Among the most active contributors on Hugging Face by community interactions (1,429+ contributions), with over 9,300 followers on the platform as of 2025–2026.

Key Relationships

  • Aran Komatsuzaki (@arankomatsuzaki) — The other influencer identified alongside Khaliq in the UCSB citation study; CTO of Teraflop.ai and former contributor to GPT-J and LAION. The two accounts are frequently cited together as the primary daily paper curators in the community.

  • Clément Delangue — Co-founder and CEO of Hugging Face; institutional relationship as Khaliq’s employer, and the executive who ultimately oversaw embedding Khaliq’s curation practice into platform infrastructure via Daily Papers.

  • Gradio team / Abubakar Abid — Khaliq’s colleagues during the Gradio period and through the acquisition; the Gradio organizational context on Hugging Face remains active and Khaliq continues to contribute to Gradio-based tooling.

  • Andrej Karpathy — Among the highest-profile followers of @_akhaliq on X; his presence in Khaliq’s follower graph is indicative of the feed’s penetration into elite ML research circles.

  • Yann LeCun — NYU professor and Meta Chief AI Scientist; another prominent follower, reflecting Khaliq’s reach across both academic and industry audiences.

  • Iain Xie Weissburg (UCSB) — Lead author of the 2024 study that formally documented Khaliq’s influence on citation patterns; Weissburg has emphasized that the intent is systemic awareness rather than individual critique.

Personal Style

Khaliq’s curation practice is deliberately high-volume and low-editorial: a typical post on X or HF pairs a paper title with a brief descriptor and, where available, a live demo link, without extended analysis. In his own words, quoted in IEEE Spectrum (2024), he sees the community as undergoing “some shift… toward finding/discovering new research and citing it through Twitter or other social media rather than conferences/reviewer scores.” His open-source projects extend the same ethos — AnyChat and AnyCoder are characterized by broad compatibility, rapid iteration, and immediate public availability rather than depth in any single model or framework. He treats Hugging Face Spaces as both a professional deployment environment and a personal laboratory, making his development process unusually transparent.

References