Virat Singh is the founder and CEO of Financial Datasets, which supplies market data to AI applications. He also maintains open-source financial agents, including Dexter, and previously worked on search and payments systems at Faire and Airbnb.[2][3]
Start with Dexter →
X Career
Before founding Financial Datasets, Singh worked as a tech lead on search recommendations at Faire and as a software engineer at Airbnb and Acorns. His company profile describes an earlier start in buy-side mergers and acquisitions. Financial Datasets joined Y Combinator’s Summer 2026 batch. Its product turns company filings, disclosures and market feeds into structured information that software agents can query, rather than requiring an analyst to interpret a terminal screen. That connects his financial applications to an underlying data problem: an agent needs source material it can retrieve and cite.[2]
Contributions
Research plans that call financial tools
Dexter turns a financial question into smaller research tasks, calls tools to obtain market information, and revisits its intermediate results before producing an answer. Its repository includes loop detection and step limits, tool-call logs, and an evaluation runner so developers can inspect the research process. The project is presented as educational and informational software, with explicit limits on trading use. Singh’s earlier Financial Agent UI explores a related interface problem: generating charts and other interface elements while an agent answers a question, using financial-data and web-search APIs.[3][5]
Turning filings into training examples
The Financial Datasets Python library creates question-and-answer examples from supplied text, PDFs or SEC filings. A developer provides source material and a language-model key; the generator produces examples with an answer and supporting context. This is a dataset-construction tool, distinct from the company’s live market-data API. Keeping that distinction helps readers choose between generating examples for model development and retrieving current company information for an application.[4]
Tools, writing & talks
Tools
Start here
Dexter
github.comA financial research agent that plans tasks, retrieves market data, checks intermediate results and exposes logs and evaluations.
Start with the README’s overview and example research flow to see how a financial agent is assembled. Running it requires Bun and the relevant model and financial-data credentials; reading the architecture needs no account.
Repository owner and maintainer[3]Original source- Financial Datasetsgithub.com
A Python generator for grounded question-and-answer examples from financial text, PDFs and company filings.
Repository owner[4]Original source - Financial Agent UIgithub.com
A generative interface that combines a financial agent with dynamically rendered UI and market-data tools.
Repository owner[5]Original source - AI Financial Agentchat.financialdatasets.ai
Investment research agent that pulls financial filings, runs analyses, and drafts memos using LLMs.
Catalog workOriginal source