Talks, on the record

Conference talks & demos

A decade of stage time across security, AI engineering, and cloud-native conferences — video agents, multimodal retrieval, and production ML systems.

For organizers

Kevin Dela Rosa is co-founder and CTO at Cloudglue. Previously a Staff ML Engineer and Tech Lead Manager at Snap, he builds video agents, multimodal retrieval systems, and production ML infrastructure.

Topics: video-aware agents, multimodal retrieval, computer vision, and production ML systems.

Contact me on LinkedIn Full background

DEF CON 33 · Recon Village · 2025

Autonomous Video Hunter: AI Agents for Real-Time OSINT

Video-agent systems for real-time OSINT: target discovery, evidence extraction, and grounded reports over live video collections.

Open on YouTube

CascadiaJS · 2025

AI That Gets the Picture: Building Video-Aware Assistants

Model Context Protocol, multimodal RAG, and TypeScript patterns for assistants that understand video.

Open on YouTube

Gen AI Zoo · Seattle · 2025

Cloudglue: Turn Videos Into Structured, LLM-Ready Data

Live demo talk: from raw video collections to searchable, agent-ready context.

Open on YouTube

KubeCon + CloudNativeCon NA · 2020

MLOps at Snapchat: Continuous Machine Learning with Kubeflow and Spinnaker

Production ML operations for training, evaluating, and deploying perception systems at Snap scale.

Open on YouTube

AWS re:Invent · 2019

Build Accurate Training Datasets with Amazon SageMaker Ground Truth

Data-labeling and training-pipeline lessons from building computer vision systems at Snap.

Open on YouTube

CascadiaFest · 2015

Adding Intelligence to JavaScript Applications

An early take on shipping machine learning inside everyday web applications.

Open on YouTube