What it does
Karya is an organisation based in Bengaluru that sells Indian-language speech datasets, transcription and translation, and evaluations of AI models. According to Forbes India, workers across the country make the data on smartphones, by reading sentences aloud, describing images and subtitling, and are paid for the work. Its website, read on 9 October 2026, says it covers "22 official Indian languages" for conversational speech datasets and lists the Government of India, Microsoft, Anthropic, Google, the Gates Foundation and OpenAI under "Trusted by".
Languages with little recorded or written data get weaker AI tools. Karya's case is that paying the people who speak those languages fills that gap and puts income in their hands. TIME reported in 2023 a worker earning about $5 an hour, and Microsoft reported in 2024 that Karya pays about $5 an hour. Forbes India reported in June 2025 that it pays at least $5 an hour. Level one work, reading sentences aloud, pays $5. Level three work, subtitling, pays $10. Some of the more serious work pays $25. Forbes India gives no rate for level two, describing an image aloud. These are Karya's own figures, and we found no independent audit of what workers are paid.
Worker counts differ by source and date. TIME reported some 30,000 people paid in 2023, and Microsoft reported more than 30,000 who had worked for Karya in 2024. A NITI Aayog showcase page of June 2025 shows "35,000+" people impacted, as submitted figures it does not endorse. Forbes India reported that it works with over 100,000 people in June 2025. Microsoft says Karya began in 2017 as a project at Microsoft Research in Bengaluru and became independent in 2021, with more than 200 nonprofit partners. TIME described it in 2023 as a nonprofit with a for-profit subsidiary, and its website footer names DAIA Tech Pvt. Ltd.
We compiled this listing from public sources. We have not used the app, and Karya has not reviewed it.
- Based in
- Bengaluru
- Made for
- Companies and government bodies that need Indian-language data; workers in villages do the tasks
Compiled by the Takṣa team from public information, last updated 9 October 2026. Sources: Karya's website, TIME on Karya, 27 July 2023, Microsoft Source Asia on Karya, 7 February 2024, Forbes India on Karya, 16 June 2025, NITI Aayog Frontier Tech profile (figures submitted by Karya).
Business
Karya sells Indian-language data and AI evaluations to organisations. Forbes India reported in June 2025 that it breaks even on its projects and does not need to raise funds. No revenue or loss figure was found.
How it earns
Karya's website lists the Government of India, Microsoft, Anthropic, Google, the Gates Foundation and OpenAI under "Trusted by", and Forbes India named the Government of India, Google, Ola, Microsoft and Zoho as clients. TIME reported that Karya says workers receive the proceeds when their data is resold. The breaking-even statement is Karya's, as Forbes India reports it. No revenue figure, price list or accounts were found. Figures from Forbes India, 16 June 2025.
Who builds it
- Manu ChopraFounder and CEO
Forbes India lists him among the founders, and TIME called him CEO in 2023.
- Safiya HusainFounder
Forbes India lists her among the founders.
- Vivek SeshadriFounder
Forbes India lists him among the founders.
Company details
- Company, per company
- DAIA Tech Pvt. Ltd., 60/1, 9th Main, 14th Cross, Off CMH Road, Bengaluru, Karnataka 560038
- Legal form, per TIME (July 2023)
- Described as a nonprofit; a grant helped its founders launch it as an independent organisation in 2021
- Origin, per Microsoft (February 2024)
- A Microsoft Research project in Bengaluru from 2017, independent from 2021
- Founded, per Forbes India (June 2025)
- 2021, by Manu Chopra, Safiya Husain and Vivek Seshadri
- Nonprofit partners, per Microsoft and TIME
- More than 200
Read on 9 October 2026 by the Takṣa team. The company has not reviewed this page, and anything that changed after that date is not shown.
How to get it
- Languages
- 22 official Indian languages for conversational speech datasets, according to its website (read 9 October 2026)
Checked 9 October 2026 · Sources · Report a correction
Why it fits India
Says it pays rural workers, reached through local nonprofits, to record and transcribe speech in Indian languages.
- Languages
- Karya's website lists conversational speech datasets in the 22 official Indian languages.
- Finding workers
- It recruits through local non-profit partners. TIME and Microsoft both put the number at more than 200.
- Pay
- Forbes India reported in 2025 that it pays at least $5 an hour, with some work paying up to $25. These are company figures.
- Resale share
- TIME reported in 2023 that workers share the revenue when a dataset is resold, and that $116,000 had been paid to about 4,000 workers by July 2023.
