AI People Wiki

Builders & open source

Avi Sil

NLP researcher; first author of IBM's open-source PrimeQA toolkit and co-author of Granite Embedding

Avi Sil (Avirup Sil on arXiv) is a natural language processing researcher whom IBM Research described in 2022 as an IBM researcher who led development of the Dr. Decr and GAAMA question-answering models behind PrimeQA. Sil is the first author of the PrimeQA paper, which introduces an open-source repository for multilingual question answering covering retrieval, reading comprehension, and question generation, released under the Apache-2.0 license. Sil is also a co-author of the 2025 Granite Embedding Models report, whose English and multilingual retrieval models IBM released under the Apache 2.0 license.12345

At a glance

Current role
None confirmed in the sources
Work listed
2
Sources cited
6
Profile reviewed

Research

Sil's papers in the ACL Anthology begin in 2011 and through the 2010s focus on information extraction, including entity linking ("Linking Named Entities to Any Database", EMNLP 2012), language-independent named entity linking (ACL 2016), and cross-lingual relation and event extraction (EMNLP 2019).6

From 2020 the work shifts toward question answering and retrieval, with papers on span selection pre-training, the TechQA dataset, a reading comprehension system covering more than 100 languages, and cross-lingual information retrieval learned from an English retriever. Later work includes CLAPnq, a long-form question answering benchmark for retrieval-augmented generation (TACL 2025), and agent research such as SMART, on reducing unnecessary tool use (ACL Findings 2025).6

Open-source work

A July 2022 IBM Research blog post describes Sil leading development of the multilingual question-answering models Dr. Decr and GAAMA and presenting Dr. Decr at NAACL. Sil is first author of the PrimeQA paper (ACL 2023 demos), which describes an Apache-2.0 repository that gathers retrieval, reading comprehension and question generation methods so researchers can reproduce and extend each other's work.13426

Work and contributions

  • IBM Granite Embedding models Open weights

    Listed among the authors (as Avirup Sil) of "Granite Embedding Models" (2025), a family of encoder-based retrieval embedding models in English and multilingual versions that the report says are publicly released under the Apache 2.0 license.5

  • PrimeQA Open-source code

    First author of the PrimeQA paper (2023), listed as Avi Sil on IBM Research's publication page and as Avirup Sil on arXiv. IBM Research's 2022 release post says Sil led development of the Dr. Decr and GAAMA models that PrimeQA is based on; the primeqa repository is under the Apache-2.0 license.3214

Timeline

  1. Published "Linking Named Entities to Any Database" at EMNLP.6
  2. Published "One for All - Towards Language Independent Named Entity Linking" at ACL.6
  3. Co-authored "The TechQA Dataset" and "Span Selection Pre-training for Question Answering" at ACL.6
  4. Co-presented the EMNLP tutorial "Multi-Domain Multilingual Question Answering".6
  5. IBM Research described Sil's work leading the Dr. Decr and GAAMA question-answering models behind PrimeQA.1
  6. Released the PrimeQA paper as first author on arXiv.3
  7. Co-authored CLAPnq, a long-form question answering benchmark for RAG systems, in TACL.6
  8. Co-authored the Granite Embedding Models report.5

Documented roles

The sources do not confirm a current role; these are the roles they document.

  • IBM researcher specializing in natural language processing (IBM Research blog, July 2022), IBM Research1

Companies in their history

  • IBM 29 people in the wiki

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

6 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Putting more knowledge at the fingertips of non-English speakers (IBM Research blog)IBM Research · Announcement · published Jul 11, 2022 · accessed Oct 21, 2026
  2. PrimeQA: The Prime Repository for State-of-the-Art Multilingual Question Answering Research and Development — IBM Research publication pageIBM Research · Official page · published Jan 25, 2023 · accessed Oct 21, 2026
  3. PrimeQA: The Prime Repository for State-of-the-Art Multilingual Question Answering Research and Development (arXiv 2301.09715)arXiv · Paper · published Jan 23, 2023 · accessed Oct 21, 2026
  4. primeqa/primeqa README and license (Apache-2.0)PrimeQA (GitHub repository) · Repository · accessed Oct 21, 2026
  5. Granite Embedding Models (arXiv 2502.20204)arXiv · Paper · published Feb 27, 2025 · accessed Oct 21, 2026
  6. Avirup Sil — ACL Anthology author pageACL Anthology · Paper · accessed Oct 21, 2026