career

The timeline.

From automating spreadsheets to managing platform deployments.

2018–2020

Customer Service Quality Manager

Amazon

Started in customer service quality — the work was repetitive but the broken processes were everywhere. I automated them with Python, SQL, and VBA, saving about 40 hours a week. Built QuickSight dashboards that turned raw numbers into decisions. Eventually owned quality strategy for operations handling 10M+ customer contacts a year, with CSAT improvements measured in basis points.

10M+ annual contacts QuickSight dashboards Python/SQL/VBA automation 40 hours/week saved
2020–2022

Service Delivery Leader, Screening Services

Amazon

Led the India background check revamp. The 90th percentile turnaround was four months — candidates would leave. Worked with Legal, Compliance, and Business to renegotiate vendor SLAs, parallelize the workflow, and build a team from 10 to 45. Cut it to one month, impacted 120K annual hires.

120K annual hires impacted 4mo → 1mo turnaround 10 → 45 FTEs Mentored 6 leaders
2023–Present

Global Payroll Payments Manager

Amazon

Define technical requirements for two engineering teams building an in-house payroll platform across 6 APAC markets, $1.2B annually. Coordinate payments integrations via API and SFTP, run fraud detection on 1.5M transactions. The job is less about building tools and more about making sure the right tool gets built in the first place.

$1.2B annual payroll 6 APAC markets 1.5M transactions analyzed 80% automation gain

The first lesson from customer service: if you can measure something, stop guessing about it. Building QuickSight dashboards that turned raw numbers into actual decisions — that was the moment I realized most arguments just end when the data loads. Not because data is always right, but because it forces you to be honest about what's happening.

The background check revamp changed how I think about process. The turnaround was four months — candidates would leave. Fixing it wasn't about code. It was getting Legal, Compliance, and Business on the same page, renegotiating vendor SLAs, and earning trust that the new workflow was safe. You have to understand the system before you can fix it.

Automation is tempting, but the real question is "should this be automated?" Python scripts saved 40 hours a week early on. The fraud detection engine reduced manual validation by 80%. But the remaining 20% — the edge cases that need judgment — that's where the work actually matters. The easy wins are easy for a reason.

The homelab isn't a side project. It's where the professional and personal overlap — 50 services on bare metal, no cloud provider, full observability, self-hosted Matrix federation, and AI inference. There's a specific satisfaction in knowing exactly where your things live and why they work. It's just stuff that works, and I like controlling where it lives.


Tools & Domains

PythonSQLQuickSightVBADockerLinuxKubernetesGrafanaPrometheusLokiCaddyMatrix
ISO 20022
PaymentsInfrastructureProcess Engineering

contact

Pick a channel.