Meet the Team

Dan Roth was founder and CEO of three previous technology companies which were all acquired by public companies, before co-founding Scaled Cognition. His first venture, started while in college was Voice Signal Technologies (VST), a pioneer and market leader in speech interfaces for mobile phones. VST was the first company to develop speaker independent voice interfaces for mobile devices in 1998 for dialing and search applications, licensing the technology to every major handset OEM globally.

VST went on to create the first ever voice-to-text technology for messaging for mobile phones, deploying it on Samsung handsets in the USA in 2002. VST was the first voice control software to be integrated into the Apple iPhone. After VST was acquired by Nuance, Dan went on to create Shaser BioScience, a pioneer in FDA cleared, OTC lasers for consumer cosmetic applications. Shaser Bio was acquired by Spectrum Brands in 2012 after shipping millions of devices worldwide.


Dan then founded Semantic Machines, together with Dan Klein to solve what is now known as "agentic" natural language processing, where neural models have the ability to write their own software in real-time (program synthesis) to call functions and take actions needed to solve user requests. Semantic Machines was a pioneer in agentic action model technology creating the first models and development environment needed to make robust language interfaces possible. Semantic Machines assembled one of the most elite teams of AI researchers ever, and in 2018 was acquired by Microsoft. Dan went on to run Conversational AI at Microsoft as Corporate Vice President for 5 years before co-founding Scaled Cognition.

Dan Klein is a serial entrepreneur and professor of AI at UC Berkeley. One of the best-known AI researchers, Professor Klein and his students have a track record of advancing the state of the art, with many of his students going on to run teams at Google, Microsoft, OpenAI and elsewhere.

As an entrepreneur, Dan founded two successful startups prior to Scaled Cognition. The first was adap.tv, an early leader in applying machine learning to advertising, acquired by AOL.

Dan went on to co-found Semantic Machines where he assembled and led an extraordinary group of researchers making breakthroughs in action taking neural models and program synthesis. After the acquisition by Microsoft, Dan was appointed a Technical Fellow for AI, where he made major contributions to Microsoft's work in AI for language interfaces.

Damon Pender, Co-Founder and CFO, has over 30 years of experience starting and growing technology companies. He is responsible for the company's finance and regulatory issues, corporate matters, human resources, legal affairs and facilities.

Prior to Scaled Cognition, Damon was Co-Founder and CFO of Semantic Machines, Inc. Semantic Machines was a pioneer in agentic action model technology creating the first models and development environment needed to make robust language interfaces possible. Semantic Machines assembled one of the most elite teams of AI researchers ever, and in 2018 was acquired by Microsoft.

Damon has also held positions with TeraDiode, Inc. (sold to Panasonic Corporation in 2017), Shaser BioScience(sold to Spectrum Brands in 2012), Voice Signal Technologies, Inc. (sold to Nuance in 2007), and PriceWaterhouse Coopers.

Adam is a research scientist at Scaled Cognition whose natural language research experience spans academia, big tech, and startups. His academic work during his PhD at Berkeley focussed on efficient learning and inference techniques for structured prediction, and won a Best Paper award at ACL in 2009. Since his PhD, Adam has built conversational assistants at Google, as one of the first employees at Semantic Machines, and at Microsoft. He is particularly interested in structured controls and guardrails for LLMs.  

Previous Affiliations:

• PhD, UC Berkeley
• Senior Research Scientist, Google
• Senior Principal Research Scientist, Microsoft Semantic Machines

Anthony is a research scientist at Scaled Cognition with over a decade of experience in artificial intelligence research, bridging fundamental methodologies and real-world applications. He received his PhD at Carnegie Mellon University, where he pioneered work in unsupervised, semi-supervised, and multi-task learning, and was awarded the CMU Presidential Fellowship. Prior to joining Scaled Cognition, Anthony led the design and development of a foundation model for anticipating and accelerating user interactions, at Microsoft. At Scaled Cognition, his research is centered on advancing LLMs beyond token-based reasoning, focusing on modeling and training techniques that enable models to reason over actions while incorporating structured controls and constraints during inference.

Previous Affiliations:

• PhD, Carnegie Mellon University
• Senior Principal Researcher, Microsoft Semantic Machines

Chris Kedzie is a research scientist at Scaled Cognition with a background in natural language generation and conversational AI. He holds a PhD from Columbia University, where he focused on controllable generation and deep-learning-based language models. Chris has contributed to popular open-source tools like Rasa Open Source and worked previously at Microsoft designing enterprise copilot evaluation. His research has been published at top conferences such as ACL, EMNLP, and AAAI, and he received a best paper award at INLG 2019. When not doing language model research, he enjoys playing electric guitar, enjoying live music, and making pizza with his wife and dog.

Kevin Yang is a research scientist at Scaled Cognition.

Prior to joining, he did his PhD in AI / NLP at UC Berkeley. His work focused on structured planning, control, and alignment/instruction-tuning methods for improving the quality and safety of natural language generation, especially for long context.

Before his PhD, he did his undergrad and master's at MIT, where he did research on using AI for pharmaceutical drug discovery.

Previous Affiliations:

• PhD, UC Berkeley
• Visiting Researcher, FAIR
• MEng, MIT

Matt is a research scientist at Scaled Cognition who has been conducting natural language processing research for nearly two decades, with close to 100 academic publications.

From his PhD at Carnegie Mellon University to his work at AllenAI, Matt’s early research focused on getting machines to read, including winning a best paper award at NAACL 2018 for ELMo, the first major precursor to modern LLMs.

Since the rise of LLMs, he has focused on how best to use them to get things done, as tech lead for the initial development of M365 Business Chat at Microsoft, and in his work building and evaluating LLM agents at Scaled Cognition.‍

Previous Affiliations:

• PhD, Carnegie Mellon University
• Senior Research Scientist, Allen Institute for AI
• Senior Principal Researcher, Microsoft Semantic Machines

‍

Mitchell is a research scientist at Scaled Cognition. Prior to joining, he completed his PhD at UC Berkeley, where he focused on building structured models and developing structured decoding algorithms for natural language processing. During his PhD, he also worked part-time at Google Brain, where he collaborated with the Transformer team to create faster and more flexible models for machine translation.

Nikita Srivatsan is a research scientist at Scaled Cognition, having joined in June of 2024. Prior to that, she earned her PhD in Language Technologies from Carnegie Mellon University, during which she published research on structured representation learning for multimodal and language-adjacent data across varied domains from typography to music. Along the way, she interned at Twitter, and also worked on a research collaboration with Boeing.

Previous Affiliations:

• PhD, Carnegie Mellon University
• Research Intern, Twitter
• Research Collaborator, Boeing

Aviv Keshet is a software engineer at Scaled Cognition, with a focus on ML systems and performance. Prior to joining Scaled Cognition, Aviv was a Staff Software Engineer in the systems architecture team at Aurora Innovation, designing and implementing safety-critical performance instrumentation and infrastructure for autonomous vehicles, and prior to that managed a test infrastructure team for ChromeOS at Google. He holds a PhD in physics from MIT, where he built a novel apparatus for atomic physics experiments and developed a widely used control software suite for such experiments. In his free time, he enjoys cooking, spending time with his family, traversing the city with them by cargo bike, and dancing argentine tango.

Brent is a software engineer with experience working on platform reliability at Square and ML infrastructure in Azure. Outside of work, he enjoys baking and is currently on a quest to find the best bagel in Boston.

Jennifer Wang is a software engineer at Scaled Cognition. Previously, she worked with the Onsite Publishing team at Amazon on data ingestion pipelines and UXs for associates. In her free time, she enjoys competitive Magic: the Gathering, as well as cooking and puzzles.

JJ Wang is a software engineer at Scaled Cognition, where she focuses on developing seamless user experiences. Previously, she led a team at Flexport, delivering operational solutions for high-demand environments, customers, and internal tooling across teams. Prior to that, she worked at Microsoft, contributing to large-scale online services, driving the shared checkout experience across products and teams.

Jordan DeLoach is a software engineer at Scaled Cognition. Previously, he served as a Staff Software Engineer at Twitter, where he led efforts to optimize the ad funnel across consumer, policy, and ads engineering organizations. Prior to that, he was at Microsoft Semantic Machines, where he led the team responsible for shipping Semantic Machines AI experiences in Outlook and Teams mobile apps.

Lisa is a software engineer at Scaled Cognition, with previous experience on the Cloud Infrastructure Capacity Management team at Salesforce.

A veteran technologist, previously at Microsoft, with 24 years building large-scale online services and pioneering security innovations in Exchange Online. In my downtime, I enjoy noodling on guitar and hanging out with my wife and cats. At Scaled Cognition, I talk to machines—the machines mostly listen.

Akshay comes to Scaled Cognition after 14 years at Google as a Group Product Manager, passionate about bringing innovative AI and ML experiences to market.

Previously, he led Gboard for Android, building industry leading autocorrection, swipe typing, and voice typing on all Android devices with new ML models and evaluation techniques. He built some of Google's first-ever on-device LLM use cases, including Smart Replies and Proofreading on Pixel phones.

Prior to Gboard, he launched new Android features including Autofill on Android, Nearby for proximity-based discovery, and the Emergency Location Service to improve emergency calling location.

Prior to becoming a PM, Akshay was a software engineer and continues to be a hacker at heart.

Diana's background includes leading product development at Microsoft Semantic Machines for Conversational AI developer tooling, and the data production pipeline. Previously, she worked as a Data and AI Consultant at Microsoft, designing and implementing data analytics solutions for enterprises. She also has a MBA from MIT Sloan School of Management.