Farbod Tavakkoli

Speaker Bio

Farbod is a Data Scientist at AT&T focused on GPU-agnostic pre-training and post-training of domain-specific large language models for large-scale deployments, including dense and sparse (mixture-of-experts) architectures. His work spans the full generative AI lifecycle, from data curation and model design to evaluation, benchmarking, deployment, and open-source enablement. 

A significant portion of his work centers on research and evaluation of language models, with contributions to modern benchmarks such as TeleLogs, Spider 2.0, and BIRD. In parallel with his industry role, Farbod is a PhD candidate at Southern Methodist University conducting applied research at the intersection of data science, urban systems, health, and policy analytics. His background spans engineering and applied research across transportation safety, healthcare, climate and environmental analytics, housing stability, and social equity.