WhatsApp infra

Working on a distributed in-memory database

Ads Creative & Ecosystem

• Worked on the Monetization Creative team, developing video audit workflows and infrastructure using Go.

• Added video appeal features to the audit flow using Eventbus async pipeline, increasing approval rate by 7%.

• Introduced music copyright detection to drive companies to join the platform, boosting revenue by $100M+.

• Built a search pipeline with Elasticsearch handling 400+ QPS, migrating data via RocketMQ and Lambda.

• Optimized 10TB+ asset library by reindexing Elasticsearch, enhancing binlog pipeline, reducing latency by 10%.

Platform infra - Anti-Abuse

• Worked on the Anti-Abuse team for HTTP traffic protection against malicious users using C++ and Lua.

• Migrated our product to cloud by Kubernetes on AWS Elastic Kubernetes Service, reducing 20% latency.

• Applied Java, Python, and wrk2 to add multi-AZ performance and failover test suites to SafeT-AWS test bed.

• Developed an application to monitor cluster and safeT configuration changes, storing them in Redis database.

• Utilized Screwdriver to build CI/CD pipelines, enhancing web application firewall and rate limiter efficiencies.

AI infra

• Worked on the Artificial Intelligence Infra team, responsible for deploying and inferring models using C++.

• Created a face detection software module and integrated it into computer vision SDK, with 100+ calls per day.

• Applied MNN and SNPE engines to deploy models on CPU and DSP chips on the ARM architecture.

• Developed tools with custom model quantization and operator introduction functions employing Flatbuffers.

• Built a functional software module that can fuse multiple TFLite Graphs, reducing 60% message transfer time.