You might catch me floating around the internet as LikeMike23
Developer Relations at Epic Games (Fortnite), 443ID (now Imprivata), and Artemis, where I also worked as a frontend dev. Before all that, six years in the United States Marine CorpsEx-VC-backed founder (round led by a16z)(Sometimes)Ex-developerCyber Security enthusiastBrazilian Jiu Jitsu blue belt (4 years)
The account-level decisions behind deploying SecureStack AI to real ECS and RDS infrastructure: a scoped IAM identity instead of root, two separate permissions boundaries to close a subtle PassRole gap, a compensating IP restriction for a longer-lived demo, and what it actually cost.
Building SecureStack AI taught me that grounding an LLM's answers is the easy part. The real work was a verification layer that checks citations, tool-call evidence, and deterministic scanners so nothing gets trusted just because it sounds confident, plus two AWS deployment gotchas along the way.
Everything you need to go from a blank machine to a working local AI setup with Ollama: installation, pulling and running models, the CLI reference, the REST API, and Modelfile customization, all in one place instead of scattered across five doc pages.
Detecting account takeover is half the problem. This one covers what Verosint actually does once a login gets flagged, from automated MFA enforcement to VERA, its agentic AI security analyst that handles triage over Slack.
A technical look at how the Verosint platform I worked on as a Developer Relations engineer detects account takeover: unified observability across identity systems, AI-driven risk scoring, and a triage layer built for investigation speed.