2026 Sarthak founds Article.One

1. I started Article.One.

   • Source: Article One — homepage

2025 Sarthak authors Section 6621 of the Intelligence Authorization Act of 2026

1. I authored Section 6621 of the Intelligence Authorization Act of 2026. Passed as part of the 2026 NDAA.

   • Source: Congress.gov — S. 1071 text

2. I drafted & supported the CLEAR and FORCE Acts of 2025 to modernize the Defense Production Act wrote proposals that unanimously passed the House Financial Services the modernization and effectiveness of the Defense Production Act. Passed House Financial Services unanimously in Spring 2026.

   • Source: U.S. House Committee on Financial Services — Markup of Various Measures
   • Source: Congress.gov — H.R. 7688 text

2024 Sarthak authors the Hawley-Blumenthal AI Risk Evaluation Act

1. I authored the Hawley-Blumenthal AI Risk Evaluation Act of 2025. My proposals were implemented broadly as made, with some adjustments for the changed policy climate in 2025. I later advised on the incorporation of some of this work in the Obernolte-Trahan FRONTIER AI Act of 2026.

   • Source: Congress.gov — S. 2938 text
   • Source: CSET — Blumenthal and Hawley’s U.S. AI Act Framework
   • Source: TechGov — The Hawley-Blumenthal AI Risk Evaluation Act

2. I advised the Center for Strategic and International Studies on their 2025 report "Advancing Federal Adoption of Cloud" on behalf of my boss Jim Himes (Cloud Commission Congressional Co-Chair). I then worked tirelessly across the Hill and with Financial Services agencies to find partners to implement these findings, with zero success.

   • Source: CSIS — Faster to the Cloud

2023 Sarthak co-designs the Senate’s landmark AI hearing series — on both the risks and the benefits.

1. I co-designed and co-organized the Senate Judiciary Subcommittee on Privacy, Law, and Technology's 2023-2024 series of AI hearings and related legislation on the risks and opportunities of the technology as it gained national prominence.

   • Source: U.S. Senate Judiciary Committee — Oversight of A.I.: Principles for Regulation
   • Source: U.S. Senate Judiciary Committee — AI and the Future of Journalism
   • Source: U.S. Senate Judiciary Committee — Oversight of AI: Election Deepfakes
   • Source: U.S. Senate Judiciary Committee — Oversight of AI: Insiders’ Perspectives
   • Source: Politico Pro — Cruz blocks Hawley’s AI bill

2022 Sarthak runs world's first investigations into election manipulation and LLM-generated disinformation at OpenAI.

1. I proposed, designed, and ran the world's first investigation into the use of frontier AI systems to manipulate democratic elections in support of OpenAI's Security & Safety teams, focusing on the US 2022 midterms. I then designed a generalizable system of detection that could be automated to a larger class of such safety issues.

2. I responded live to the world's first analysis of AI generated misinformation in the wild at scale for the OpenAI Safety Operations team. After documenting it live for 96 hours straight, I built the first systems to reduce the risk of this abuse from recurring.

2021 Sarthak builds systems for LLM Monitoring, Safety Reviews, and fighting child abuse at OpenAI.

1. I built the world's first monitoring system for usage of LLM systems by consumers. I led OpenAI to allow doubling the generation limit from 2048 to 4096 tokens per response, by convincing leadership that doing so would not lead to the creation of bioweapons.

   • Source: OpenAI Developer Community — token-limit discussion

2. I developed and supervised OpenAI's "Safety Review Process" to identify which developers would be allowed to use GPT-3 based on their proposed safety mitigations and procedures. Companies did not have catastrophic safety incidents under my oversight.

   Note: No news is good news.

3. I was the product manager responsible for OpenAI's first systems to respond to the production of text-based 'child pornography' by users of GPT3.

   • Source: WIRED — The AI-fueled dungeon game got much darker

2020 Sarthak starts fully funded graduate program in pure mathematics at University of Connecticut.

1. I started a fully funded graduate program in pure mathematics.

   • Source: OpenPayrolls — Sarthak Agrawal

2019 Sarthak co-founds Redo, fails miserably & supports early R&D for TypeScript validation engine ArkType.

1. I co-founded Redo, a test automation company that failed miserably. I supported the early R&D for ArkType, the world's premier validation engine for Typescript.

   • Source: ArkType — project website

2018 Sarthak does early research in hyperparameter selection at state-of-the-art NLP/AI startup & graduates with B.A. in South Asian Languages and Civilizations.

1. I built production grade data engineering systems at Quillbot and did early research on hyperparameter selection to improve their AI systems. The systems beat state of the art before my work, and did better after.

   • Source: QuillBot — About

2. I graduated from the University of Chicago with a Bachelor of Arts in SALC (South Asian Languages and Civilizations). I studied devotional poetry & religious texts from South Asian religious traditions in Sanskrit, Hindi, and Braj.

2017 Sarthak realizes Machine Learning (ML) will destroy finance careers & learns to code.

1. I was hired to build a system to accelerate debt investment decisions. I read "Elements of Statistical Learning", was smart enough to know I was out of my league, and honest enough to know Machine Learning (ML) was going to destroy my career.

2. I taught myself to code so I could switch fields into AI.

   • Source: Article One — Director CV

2016 Sarthak learns to invest at Bridgewater Associates & designs novel investment strategies.

1. As an intern at Bridgewater Associates, I studied how they invested institutional capital. I then used legal research techniques to adapt these into novel strategies applicable for UHNW (Ultra High Net Worth) individuals.

2. At my mid-point meeting, I told the firm's best investors that "I fucking hate my life here." When they apologized and offered to let me go with (1) my signing bonus, (2) the rest of my summer pay, and (3) excellent references, I said no. I am less of an idiot now.