Turning repeat IT work into support people can use.

At Block, I worked across identity, devices, SaaS, and executive AV for a global team. I automated onboarding, built the knowledge system, and co-built an AI support assistant so people could solve common issues without waiting.

ROLE
IT Support Engineer, L4 · Endpoint & Device Management
DATES
Jan 2020 – Apr 2025
BASE
Tokyo, Japan
COVERAGE
North America · LATAM · Japan · APAC · India · EMEA
200
new hires per week
at peak, onboarded through automated provisioning
~35%
fewer incoming tickets
measured after the knowledge base landed
300+
technical articles
written and maintained in Confluence
~20%
IT tickets deflected
by the RAG support assistant

Support that holds up when the stakes are high.

I supported CEO Jack Dorsey, C-suite executives, and board members directly. I prepared and ran AV for company-wide all-hands and board meetings, alongside the device and account work that keeps a global team moving.

I arrived early, tested every input twice, kept a spare of the likely failure points, and gave each person the amount of technical detail they needed to start. That same discipline shaped the automation and documentation work around it.

The goal was simple: give people a working answer before the meeting starts.

A checked onboarding path for up to 200 people a week.

I connected the accepted-offer workflow to a working, compliant device, then used the same controls to close access and recover assets when someone left.

WHAT I BUILT
  • Account and group provisioning through Okta and Azure AD / Intune
  • Google Workspace and Microsoft 365 setup per role
  • Device enrollment and compliance baselines through MDM
  • Deprovisioning and secure asset recovery on exit
WHAT IT CHANGED

I automated provisioning so first-day IT followed a checked path, even when forty people started at once. Offboarding used the same rules instead of a separate scramble.

I worked with Security, People Operations, Engineering and Facilities so access, deprovisioning, and compliance checks followed the same rules for every team.

300+ articles that move a ticket forward.

I wrote for the person already in the middle of a problem: one task per article, exact screen names, the error text they saw, and the next step if it failed. Every repeated answer became a candidate for documentation, and articles that kept generating tickets were rewritten.

Jira ITSM

Managed all incidents, requests and changes; refined workflows and reported on SLA performance.

Asset lifecycle

Inventory, procurement, repairs, replacements and secure recovery through Workspace ONE and Oomnitza, with vendors and internal logistics.

On-call

Priority rotations covering identity, endpoint compliance, AV during executive events and SaaS outages, with clear status updates and timelines.

An AI support layer built on trusted knowledge.

I co-built a retrieval-augmented Slack bot and web app over the 300+ article Confluence knowledge base. It answered IT questions in Slack and triaged Jira tickets using the material the team already maintained.

I wrote the knowledge content and the Slack and Jira integrations, then tuned answer quality. The assistant deflected about 20% of IT tickets and gave the team 24/7 first-line coverage.

The knowledge base became a first response people could reach from Slack.

Support that gets better with every solved problem.

The useful pattern is straightforward: capture the answer, connect it to the workflow, and make it easier to find the next time. That is how I help teams reduce repeat questions without hiding the detail people need.