Darrius Taylor

Product Builder · Agentic AI Systems · Security & Trust UX

askdarrius@gmail.com · linkedin.com/in/darriustaylor ·Austin, TX

I started as a software engineer and never stopped shipping. Four years designing Google Cloud's bot-defense and access-control products, and building the multi-agent AI systems my design org now runs on, architecture through deployment.

  • Built and shipped a multi-agent AI orchestrator (Node.js and MCP-server backend on Cloud Run, injected-script delivery, JSON schemas for human/agent interchange) that runs heuristic evaluation against live coded prototypes. Now the default workflow for a Google Cloud security design org.
  • Four years designing probabilistic verdicts: what a machine-versus-human judgment should look like to a customer who has to act on it. Separated verified from merely unique automated agents rather than shipping one ambiguous count, and chose the chart engineering could stand behind over the more impressive one.
  • Moved a customer-blocking diagnostic onto the Google Cloud roadmap a full year ahead of plan by building a working prototype before it was funded and running the enterprise research that made the case.

Google · Cloud Security

Jan 2022 – Present · Austin, TX

Senior Interaction Designer, Cloud Platform Security (Nov 2024 – Present) · Interaction Designer, Cloud Security (Jan 2022 – Nov 2024)

Agentic Security & Abuse Defense

  • Interaction designer for reCAPTCHA: the admin console, the policy-authoring experience, and the analytics surface where a customer sees how much of their traffic is automated.
  • Defined how automated-agent traffic is counted and shown (verified versus merely unique), so the number an admin acts on is one engineering can defend.
  • Designed the risk-simulation experience for a preemptive security product: side-by-side before-and-after exposure scores, and pending states that hold a user through a 24-hour verification window without reading as failure.
  • Replaced a multi-screen, expert-only configuration flow with a conversational canvas validated in usability testing, collapsing setup to a single guided step.

AI Systems I Built

  • Multi-agent orchestrator: owned MVP scope, architecture, and the CI/CD work that kept it deployable. Re-architected from browser extension to injected script when the extension model could not reach every platform.
  • Live commenting infrastructure for coded prototypes: defined the JSON schemas for human and agent interactions and stabilized the backend deployments.
  • Design-system delivery layer with model-readable component instructions, so generative tools produce real components rather than approximations.

Platform & Enterprise Security

  • Led end-to-end UX for the second generation of the VPC Service Controls troubleshooter, the tool customers use to diagnose why a request was blocked by a security perimeter. Built the prototype before the work was funded and ran research with large financial-services and payments customers to make the case, moving it onto the roadmap a full year early.
  • Made the case for a reusable policy-authoring primitive after finding every Google Cloud security surface had built its own, and socialized it until teams outside my own adopted it; it became the foundation for a major access-control feature.
  • Led UX and research strategy for a cross-team sign-in consolidation. When research support changed mid-project, stood up a substitute methodology (proxy A/B, desk research, internal customer-engineer sessions) and wrote the screeners, scripts, and insights schema the team still runs.

Capital One

Apr 2017 – Dec 2021 · McLean, VA

Principal Creative Technologist (2021) · Creative Technologist / Senior UX Designer (2019–2020) · Senior Software Engineer (2017–2019)

  • Shipped production front-end code for consumer banking, then moved into design and kept shipping it, carrying concepts from interaction design through implementation rather than handing off a spec.
  • Settled direction with stakeholders using working prototypes instead of mockups, so decisions were made against something real before engineering committed.
  • Selected to a company-wide AI practice cohort; helped define the discipline's hard-skill floor, arguing designers need version control and deployment rather than only better prototyping tools.
  • Co-led AI prototyping workshops for a Google Cloud security design team and wrote the onboarding and deployment guides that took designers from zero to shipping their own working prototypes.
  • Instituted the asynchronous pre-launch bug-hunt process design and engineering now run each release, and drove the accessibility backlog to zero blocking WCAG AA contrast failures before launch.
  • Drafted the product-experience objectives and quarterly OKRs for a security product line, and the growth strategy behind its console revamp: demonstrate value before asking for billing.
Build
React, TypeScript, HTML/CSS · Node.js, MCP servers, Cloud Run · JSON schema design · CI/CD
Design
Interaction and product design · design systems · agent UX, trust calibration, model-error and correction flows · Figma · usability testing, concurrent think-aloud, proxy A/B · WCAG 2.1 AA
Domain
Bot and abuse defense · access control and policy authoring · enterprise identity · fraud and risk signals