Volunteers helping people at a humanitarian aid center
Humanitarian

Tarjimly Connects Refugees to Interpreters in Minutes Using AI

7 min read|Updated September 2026
Share

At a glance

Tarjimly is a nonprofit app that uses machine learning matching to connect refugees and aid workers with volunteer translators in minutes over text, voice, or video. Its network now covers more than 250 language pairs, serving a world where over 120 million people are forcibly displaced and language barriers can decide medical and asylum outcomes.

  • Tarjimly was founded in 2017 in response to the Syrian refugee crisis
  • Its volunteer network now covers more than 250 language pairs, up from around 80 at launch
  • A 2024 partnership with Google added large language models for instant translation needs
  • More than 120 million people are forcibly displaced worldwide according to UN estimates
  • Tarjimly hires and trains bilingual refugees as paid professional translators for sensitive scenarios

Tarjimly is a nonprofit app that connects refugees, immigrants, and aid workers with volunteer translators and interpreters in minutes, using machine learning to find the right person from a global volunteer network, and it matters because language barriers are one of the most immediate dangers displaced people face. A refugee who cannot communicate with a doctor may not understand a diagnosis or a dosage. A family filling out an asylum application in a language they cannot read risks errors that can sink their case. Professional interpreters exist but are scarce, expensive, and rare in the languages that refugees actually speak.

Founded in 2017 in response to the Syrian refugee crisis and backed by Google, Tarjimly has been described as Uber for translators. A person needing language help opens the app, requests support, and an algorithm matches them with a qualified volunteer anywhere in the world who can assist over text, voice, or video.

Matching Is the Hard Part

The core technical problem is not translation itself but matching. When Tarjimly launched its app in 2019, its volunteer network collectively spoke around 80 languages, and the machine learning system had to decide, in real time, which volunteer to alert for each request, weighing language pair, dialect, availability, ratings, and the sensitivity of the situation. The organization says its network now covers more than 250 language pairs.

Dialect matters enormously in this work. A volunteer fluent in formal Arabic may struggle with a Syrian or Sudanese dialect. The matching system learns from feedback after each session, improving its ability to route requests to volunteers who have succeeded with similar needs before.

Where AI Fits In

In 2024, Tarjimly deepened its use of artificial intelligence through a partnership with Google. As the Associated Press reported, the collaboration lets Tarjimly use large language models to assist with translation needs while continuing to train its systems, allowing volunteers to respond more urgently to requests. The design is deliberately layered: instant machine translation handles quick, low stakes needs, while human volunteers remain the backbone for medical consultations, legal appointments, and asylum interviews where nuance and trust are critical.

This hybrid approach reflects a hard won lesson of humanitarian technology. Off the shelf machine translation often performs poorly in the less serviced languages refugees speak, and errors in a medical or legal context can cause real harm. Keeping humans in the loop, with AI speeding the connection and supporting the conversation, captures the speed of software without betting lives on generic models.

Refugees Helping Refugees

One of Tarjimly's most distinctive choices is hiring refugees themselves. For rarer languages and sensitive scenarios, the organization pays and trains bilingual refugees as professional translators and interpreters, creating what it describes as a feedback loop of economic empowerment. The people best positioned to interpret for displaced communities are often members of those communities.

The volunteer side scales the model globally. Bilingual professionals anywhere can donate a spare half hour from a phone, and thousands have, turning the world's diaspora populations into an on demand interpreting workforce.

Language Access as Aid

With more than 120 million people forcibly displaced worldwide according to UN estimates, the need is structural. Displacement almost always crosses a language border, and every service a displaced person needs, shelter, healthcare, legal status, schooling, work, runs through conversation. Organizations from the UN refugee agency to local clinics have identified language access as a persistent, underfunded gap in humanitarian response.

Tarjimly's answer is modest in cost and large in consequence: an app, a matching algorithm, a corps of volunteers, and now AI to make the connection nearly instant. The next time a doctor in a Greek camp needs to explain a treatment to a Kurdish speaker, or a lawyer in Texas needs to prep an Afghan client for an asylum hearing, the interpreter may be one tap away, somewhere across the world, waiting to help.

Common Questions

How fast can Tarjimly connect a refugee to an interpreter?

Requests are typically matched in minutes. A person needing language help opens the app, requests support, and a machine learning algorithm matches them with a qualified volunteer anywhere in the world who can assist over text, voice, or video.

How does Tarjimly use AI in its translation service?

Machine learning powers the matching system, weighing language pair, dialect, availability, ratings, and situation sensitivity. Through a 2024 partnership with Google, Tarjimly also uses large language models for instant, low stakes translation needs while human volunteers handle medical, legal, and asylum conversations.

Why does Tarjimly keep human volunteers instead of using only machine translation?

Off the shelf machine translation often performs poorly in the less serviced languages refugees speak, and errors in a medical or legal context can cause real harm. The hybrid design captures the speed of software without betting lives on generic models.

Can refugees work for Tarjimly as translators themselves?

Yes. For rarer languages and sensitive scenarios, Tarjimly pays and trains bilingual refugees as professional translators and interpreters, creating a feedback loop of economic empowerment within displaced communities.

Sources: Tarjimly about pages on AI/ML matching and 250+ language pairs; VentureBeat coverage of Tarjimly's 2019 launch and 80 language volunteer network; Associated Press (2024) on the Google AI partnership; UNHCR global displacement figures.