Engineering global scale multi region systems on microservice architecture - AWS, Linux, VPC, EC2, Docker, JVM, Java, Kotlin, Spring, Netty, Redis, RPC, DynamoDB, S3 - high traffic, low latency and 99.99% availability.

Develop proof-based operations to enable lock on profiles. Full stack support from backend to iOS.

Improve team productivity by broadening usage of infrastructure-as-code(Cloud Formation, CDK, Git), pipelines(CI/CD), auto-scaling, Resilience4j and CloudWatch.

PV designated auth mechanism and systems that brings more resiliency to PV traffic, tightens security by granting only essential permissions for playback, enables emergency playback access in case of unavailability of systems.

Develop and improve test strategy and libraries - unit-tests, component tests and integration tests(e2e tests), chaos tests

Introduce cross regional resiliency to core traffic shape to prevent regional failure and improve availability. (Route53, network load balancer, Redis cache, cache invalidation)

Built asynchronous I/O based cache client library that enables disaster recovery(fail-open/close), negative result caching, backward compatible serialization, and cache eviction. (Kotlin coroutine, serialization, edge caching, metrics, failure mode)

Designed avatar selection system that provides dynamic reloading, localization strings and images using AWS App-config.

Built message queue(mq) based asynchronous data processing systems. (Route53, Gateway, SQS/SNS, Lambda)

Built a data migration system to support profile level settings. Migrated 5B of PrimeVideo settings without downtime.

Participating in scrum leader rota, running meetings, collaborating with various teams in/outside of PV - front/backend dev teams, UI/UX designers and products/engineering managers. (Agile)

Horizontally scalable serving system of machine­ learned models for AI service using container­orchestration on GPU devices.

Improved performances - memory efficiency, time complexity. (n square to constant)

Developed polygon clustering (disjoint-set, dfs, convex hull algorithm)

Developed cloud­native server using Nvidia­Docker, Kubernetes, Marathon, gRPC, Python, Flask, asyncio and TF­Serving.

High availability and fault ­tolerant system, serving billions of documents data responding over 900 QPS. (avg latency: 0.3s)

Improved performances of indexing process more than 2x. (Spring, Java and search engine, POS ­tagger of Kakao)

Developed business features such as documents/keywords clustering, stable multi­thread data and request handling. (Go­lang)

Built CI/CD pipeline of entire project using Jenkins, unit­test, coverage, Docker, AWS and Git. (DevOps)

Integrated external REST spec apis into solution. (Swagger, RAML, etc.)

Developed earth scale RESTful backend using features of Geo­Hash and LevelDB. (avg latency: 0.035s)

Built infinitely scale­outable backend system with CI/CD pipeline using Docker on AWS env. (EC2, Lambda, S3, Travis.CI)