- Director, Technical Product Management, Granite, IBM Research1
Kate Soule is a Director at IBM Research who leads Technical Product Management for Granite, IBM's family of large language models, according to the IBM Research profile. Before this role, Soule was a leader at the MIT-IBM Watson AI Lab, where Soule ran the lab's corporate membership program. Soule is a co-author of the Granite Embedding Models report (2025), whose English and multilingual retrieval models are released under the Apache 2.0 license, and of papers on enterprise benchmarks for LLM evaluation and on routing among LLMs using benchmark datasets.12
At a glance
- Current role
- Director, Technical Product Management, Granite
- Company
- IBM
- Sector
- Company
- Work listed
- 2
- Sources cited
- 8
- Profile reviewed
Career
Soule's IBM Research biography describes a career at the intersection of business and AI. Soule previously worked as a senior consultant in Deloitte Consulting's Strategy and Analytics practice, and before the current role was a leader at the MIT-IBM Watson AI Lab, where Soule ran the lab's corporate membership program.1
As a Director at IBM Research, Soule now leads Technical Product Management for Granite, IBM's family of large language models. Soule also explains generative AI through IBM videos such as "What are Generative AI Models?", courses on IBM's AI Academy and regular appearances on IBM's "Mixture of Experts" podcast.1
Research
Soule's papers concern evaluating and choosing language models. A 2023 study repurposed benchmark datasets to learn a router model that selects an LLM for a new task, and a 2024 paper proposed enterprise benchmarks for evaluating LLMs on domain tasks.34
Soule is also listed on technical reports for IBM's open Granite models, including Granite Vision, Granite Embedding and Granite-speech (all 2025), and co-authored "Granite.Trust Policy Tools" (2026), which describes a YAML-based schema for specifying safety policies for generative AI applications.5267
Work and contributions
IBM Granite Open weights
Leads Technical Product Management for Granite at IBM Research, and is listed among the authors of "Granite Embedding Models" (2025), which publicly releases its English and multilingual embedding models under the Apache 2.0 license.12
Granite Switch 4.1 8B Open weights
First-listed author of the Hugging Face community article announcing Granite Switch 4.1, the release that includes the Granite-Switch-4.1-8B model.8
Timeline
- Co-authored "Large Language Model Routing with Benchmark Datasets".3
- Co-authored "Enterprise Benchmarks for Large Language Model Evaluation".4
- Listed among the authors of the Granite Vision technical report.5
- Co-authored the Granite Embedding Models report.2
- Co-authored the Granite-speech report on open-source speech-aware language models.6
- Co-authored "Granite.Trust Policy Tools" on shareable, actionable policies for generative AI applications.7
Other documented roles
Roles the sources document that are not confirmed as current.
- Ran the lab's corporate membership program, MIT-IBM Watson AI Lab1
Where they work
- IBM 29 people in the wiki
Related people
- Marina DanilevskyAlso at IBM
- Ruchir PuriAlso at IBM
- Ranit AharonovAlso at IBM
- Hendrik StrobeltAlso at IBM
- Piero MolinoAlso at IBM
- Yacine JerniteLanguage models · Evaluation and benchmarks
- Yulia TsvetkovLanguage models · Evaluation and benchmarks
- Akshita BhagiaLanguage models · Evaluation and benchmarks
- Aylin CaliskanEvaluation and benchmarks · Language models
- Collin BurnsLanguage models · Evaluation and benchmarks
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
8 sources · profile reviewed Oct 21, 2026 · data updated Oct 22, 2026
- Kate Soule — IBM Research
- Granite Embedding Models (arXiv 2502.20204)
- Large Language Model Routing with Benchmark Datasets (arXiv 2309.15789)
- Enterprise Benchmarks for Large Language Model Evaluation (arXiv 2410.12857)
- Granite Vision: a lightweight, open-source multimodal model for enterprise Intelligence (arXiv 2502.09927)
- Granite-speech: open-source speech-aware LLMs with strong English ASR capabilities (arXiv 2505.08699)
- Granite.Trust Policy Tools: Shareable, Actionable Policies for Generative AI Applications (arXiv 2608.23870)
- Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint (Hugging Face blog)