Leveraging unstructured data across HubSpot products to deliver intelligent, proactive, and impactful AI product experiences. Ensuring that SMBs have industry-leading AI tools to help them grow.

Frame AI was acquired by HubSpot in January 2025.

(https://ir.hubspot.com/news-releases/news-release-details/hubspot-acquire-frame-ai-unlocking-power-conversational-data-ai)

Frame AI is the leading Streaming Generative AI Platform for enterprise companies. Frame AI helps major brands and software companies transform natural language into proactive insights, working within existing systems and infrastructure. Frame AI’s platform leverages Stream-Trigger Augmented Generation (STAG), a breakthrough architecture designed to apply Generative AI to massive volumes of streaming data. Led by a team of PhDs in machine learning and data science, Frame AI empowers marketing, sales, CX, and product teams with tailored AI strategies.

Partners include Twilio Inc. Investors include Velvet Sea Ventures, FirstMark, Greycroft, G20, and Techammer.

Selected by senior management to lead a crop of Knewton's top engineering talent across two teams within the Adaptive Learning department. One team built core statistical inference systems, and the other built recommendation system and pedagogy QA tools. I remained a hands-on manager through periodic technical contributions of code, code review, and design review, delivering significant contributions to statistical model performance monitoring. My teams were highly autonomous and high performing due to continuously adapting process around shared accountability, quality, and business objectives. Prior to shifting into management, I built several core model and infrastructure components powering Knewton's adaptive learning platform.

Developed computer vision software for low latency human motion tracking using GPU kernels and multithreaded CPU code. Improved motion capture system user interface and remote administration tools. Built software license system and internal metrics repository comparing tracking quality and accuracy over time. Developed camera synchronization software for SLR-based 3d scanner and a 40-camera gait analysis prototype implemented as a distributed system. Designed and built a portable motion capture stage on a $40k budget designed to pack into shipping crates and featuring 12 modular light boxes fitted with fluorescent tubes, ballasts, and dimmers.

Designed and lead the manufacture of PCB and microcontroller software to synchronize an array of high-power, near-infrared SMT LEDs with camera shutter trigger signal. Designed and led the manufacture of cast aluminum camera housings and machined aluminum camera calibration wand handles fitted with off-the-shelf flashlight switches and 360-degree FOV LEDs. Developed Windows Embedded system images for Vision Processor machines and a custom Knoppix live image used for onsite updates of system software. Implemented an embedded menu system for Vision Processor front panel LCD screens.