SubjectFraud · KYC · AML · Data

RithwikManjunath

I spent three years at Amazon deciding who was safe to let in. Now I do it with data.

01

Summary of findings

Reviewed by: you
A

Has worked the case queue himself. Investigated flagged orders, payments and accounts in Amazon's Transaction Risk Management Services, then 80 to 90 KYC, AML and onboarding cases a day across three regions.

B

Fixes the process, not just the case. Spent 8 months co-designing an onboarding decision tool that went live across the organisation, and rewrote the SOPs that were causing errors.

C

Brings the analytics with him. MSc in Data Analytics, Microsoft Certified Fabric Analytics Engineer, and he uses Looker forecasting every week to run a restaurant's numbers. In his own time he built AlertIQ, an AML alert triage engine.

02

Screening hits

4 positive matches
Hit 01Volume
80–90

KYC, AML and onboarding cases closed per day at Amazon

Hit 02Accuracy
+25%

Improvement in fraud detection accuracy from spotting risk patterns

Hit 03Consistency
6/6

Months in a row as the team's top performer

Hit 04Data
−15%

Stock variance at Nando's through forecast-led ordering

03

Activity history

Most recent first
Sep 2025to present
Dublin

Operations & Performance Analyst

Nando's UK & IRE
  • Reviews sales, labour cost, stock variance, NPS and kitchen throughput in Looker daily and weekly, and traces red metrics to a root cause.
  • Uses Looker forecasts and rolling sales history to plan labour and stock, cutting stock variance by 15%.
  • Reports actuals against budget to the Regional Manager every week, with variance commentary and next actions.
Aug 2023to Sep 2025
Dublin

Supervisor

Nando's UK & IRE
  • Ran shifts across front of house and kitchen, using live station KPIs to fix problems as they happened.
  • Main escalation point for complaints, keeping NPS at 62 to 65 and Food Scores above 80.
Mar 2020to Dec 2022
Bengaluru

Registration & Compliance Operations AnalystSME

Amazon
  • 80 to 90 KYC, AML and seller onboarding cases a day across North America, Europe and Asia-Pacific, from triage to final decision.
  • Enhanced Due Diligence on high-risk merchants, business and UBO verification, and sanctions and PEP screening reviews.
  • Owned the Amazon Payments onboarding queue for MTurk and NAAPAY.
  • Co-designed an onboarding decision tool over 8 months: requirements, UAT and defect tracking with engineers until it went live organisation-wide.
  • Subject Matter Expert for the RCO BAP process: rebuilt training, audited case quality and rewrote SOPs.
Sep 2019to Mar 2020
Bengaluru

Transaction Risk Investigator

Amazon · Transaction Risk Management Services
  • Investigated orders, payments and accounts flagged by risk models, including account takeover, stolen cards and links to closed fraudulent accounts.
  • Escalated emerging fraud patterns with case examples so risk rules could be updated.
  • Contacted buyers and sellers for verification documents before reinstating or closing accounts.
04

Supporting evidence

Built in his own time · synthetic data
Exhibit AAML triage

AlertIQ

An AML alert triage engine and investigation workspace. It ranks alerts by their likelihood of becoming a SAR. On held-out synthetic data, reviewing the top 20% of the queue caught 100% of true SARs, against 20% for random order (AUC-ROC 0.979).

PythonGradient boostingNext.jsCloud Run663 tests
github.com/Rithwik23M/AlertIQ
Exhibit BFraud scoring · in progress

Fraud Risk Scoring

A model that scores mobile payment transactions by fraud risk and prioritises the analyst review queue, built on about 6.3 million PaySim transactions with a fraud rate under 0.2%.

scikit-learnSHAPSQLPower BI
github.com/Rithwik23M/fraud-risk-scoring
Exhibit CSecurity risk · in progress

HarbourPay Risk Assessment

A security risk assessment for a fictional Irish payments startup facing a partner-bank due diligence review. The goal is a scored risk register tied to a STRIDE threat model and a NIST CSF 2.0 gap analysis. So far the company profile, scoring method and automated data checks are in place.

NIST CSF 2.0STRIDEPython CI checks
github.com/Rithwik23M/harbourpay-security-risk-assessment
Exhibit DGoogle Cybersecurity

Security Audit & Incident Report

Two graded course activities, both scored 100%: a controls and compliance audit of a fictional toy company against PCI DSS, GDPR and SOC, and an ICMP flood incident report written up using the NIST Cybersecurity Framework.

PCI DSSGDPRNIST CSF
05

Capabilities

Financial crime

  • AML & KYC / CDD
  • Enhanced Due Diligence
  • Sanctions & PEP screening
  • Fraud detection
  • UBO verification
  • NIST CSF & security audits

Data

  • SQL
  • Python
  • Power BI
  • Looker & Tableau
  • Microsoft Fabric

Process

  • Requirements & UAT
  • SOP writing
  • Quality auditing
  • KPI reporting
  • Forecasting
06

Documents on file

Education & certification
Fabric Analytics Engineer Associate (DP-600)Microsoft2026✓ Certified
Google Cybersecurity Professional CertificateGoogle · 4 of 8 courses complete2026In progress
Operational Analysis of Suspicious Transaction ReportsBasel Institute on Governance2026✓ Completed
Azure Fundamentals (AZ-900)DataCamp2026✓ Completed
Business Analysis & Process ManagementCoursera2026✓ Completed
MSc Data AnalyticsDublin Business School2024✓ Awarded
Executive PG Programme in Data ScienceIIIT Bangalore2023✓ Awarded
BE MechatronicsAcharya Institute of Technology, Bengaluru2017✓ Awarded
07 · Analyst decision

Approve. Connect.

If you are hiring for fraud, risk, compliance or data analytics roles in fintech, banking or Trust & Safety, the next step is a conversation.

Connect on LinkedIn →
Self-prepared portfolio. All figures are from my own work history. Rithwik Manjunath · 2026