• Antifraud Architecture

• Feature Store

• Machine Learning

• AI Agents for Infra Automation

• AI Agents for Antifraud Risk Analysis

• Feature Store

• Risk Management

• Machine Learning Infra

• Led, designed, implemented, and launched a VOC/VOB (Voice of Customer/Voice of Business) analytics platform for TikTok Shop from zero to one, transforming raw data into actionable insights. The platform integrates the various latency data (Hive, Kafka etc.), seamlessly normalizes and stores the data (ClickHouse, Redis), generates the OLAP dashboard. It also notifies POCs (Point of Contact) in event of business incidents;

• Stakeholders from 10+ teams utilize the platform to dissect TikTok Shop customer experience, initiate proactive measures, and track the progress of resolutions;

• Enhanced operational efficiency by integrating a Large Language Model to automate customer support issue summarization for data-driven business decisions;

• Expanded the engineering team from 4 to more than 10 engineers and facilitated the onboarding of 5+ new hires; set up best practices for Software Development Life Cycle (SDLC); devised the roadmap, and successfully launched critical feature loops.

• Hired as a tech lead to build a new platform to improve third-party data ingestion on Nov/28;

• Impacted by DoorDash layoff event on Nov/30.

• Led, designed, implemented, and launched the Automated Collection of Evidence (ACE) service as a single source of truth for transportation risk compliance org. The service is architected as four layers: ingestion (SQS, Lambda), transformation (Airflow), storage (S3, DynamoDB, Aurora), and distribution (ECS, GraphQL). It helps transportation case auditing time reduced from 15 min per audit to about 1 minute per audit;

• Designed, led, and implemented a service to collect and aggregate PII (Personal Identifiable Information) data from multiple data sources; set up CI/CD pipeline, dashboard, alarms, operation runbook with SOP (Standard Operating Procedure), and compliance docs to maintain the service to continuously comply the data with FTC SLA;

• Reduced the service cost by about 40%, resolved tech debts, and attended weekly operation meetings as Ops Lead;

• Migrated GMRA (Gather, Model, Rules, Actions) workflows of merchant customers from dedicated services to cloud native machine learning platforms (AWS Sagemaker).