ksmin23/my-aws-journey-2019-2025.md

Last active December 9, 2025 01:57

Star (4) You must be signed in to star a gist
Fork (0) You must be signed in to fork a gist

Select an option

Learn more about clone URLs
Clone this repository at <script src="https://gist.github.com/ksmin23/9f797cb7a227f8476c8f1695120bb326.js"></script>
Save ksmin23/9f797cb7a227f8476c8f1695120bb326 to your computer and use it in GitHub Desktop.

Download ZIP

Raw

my-aws-journey-2019-2025.md

A Curated Collection of My Work at AWS (2019 ~ 2025)

This document curates my professional activities during my 5 years and 10 months at AWS (from August 2019 to May 2025), including blog posts, presentation decks, YouTube videos, and open-source GitHub projects. The content primarily focuses on cloud computing with AWS, covering topics such as Generative AI, LLMs, RAG, Data Analytics, Machine Learning, MLOps, and System Architecture.

Table of Contents

Blog
Speaker Deck
YouTube
Open-Source GitHub Projects

Blog

Generative AI & LLM (8)

Data Analytics (4)

Machine Learning (2)

Architecture (1)

How Channel Talk Efficiently Handles Spike Traffic with Amazon SQS / 채널톡의 Amazon SQS를 이용한 효율적인 Spike 트래픽 처리 방법 (2024-05-20) 👥

Speaker Deck

Generative AI & LLM (5)

Data Analytics (10)

Machine Learning (12)

Architecture (2)

YouTube

Generative AI & LLM (7)

Data Analytics (11)

Machine Learning (15)

Architecture (1)

7 Effective AWS / AWS를 잘 사용하기 위한 7가지 사용 팁 (2023-11-21)

Open-Source GitHub Projects

Generative AI & RAG (18)

deploy-langfuse-on-ecs-with-fargate: Provides AWS CDK code to deploy Langfuse, an open-source LLM engineering platform, on Amazon ECS with AWS Fargate.
ocr-with-amazon-bedrock: This project is an OCR (Optical Character Recognition) web application that uses Amazon Bedrock for its character recognition capabilities. The application is deployed on Amazon ECS Fargate and the infrastructure is provisioned using the AWS CDK.
rag-with-amazon-bedrock-and-kendra: A Question Answering application built with a Retrieval Augmented Generation (RAG) approach, using Amazon Bedrock and Amazon Kendra as the knowledge base.
rag-with-amazon-bedrock-and-documentdb: A RAG-based Question Answering application using Large Language Models (LLMs) powered by Amazon Bedrock, with Amazon DocumentDB serving as the vector store.
rag-with-amazon-bedrock-and-memorydb: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Amazon MemoryDB for Redis as the vector store.
rag-with-amazon-bedrock-and-opensearch-serverless: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Amazon OpenSearch Serverless as the vector store.
rag-with-amazon-bedrock-and-opensearch: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Amazon OpenSearch Service as the vector store.
rag-with-amazon-bedrock-and-postgresql-using-pgvector: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Amazon Aurora PostgreSQL and pgvector as the vector store.
rag-with-amazon-documentdb-and-sagemaker: A RAG-based Question Answering application using LLMs with Amazon SageMaker for model hosting and Amazon DocumentDB as the knowledge base.
rag-with-amazon-kendra-and-sagemaker: A RAG-based Question Answering application using LLMs with Amazon SageMaker for model hosting and Amazon Kendra as the knowledge base.
rag-with-amazon-memorydb-and-sagemaker: A RAG-based Question Answering application using LLMs with Amazon SageMaker for model hosting and Amazon MemoryDB for Redis as the vector store.
rag-with-amazon-opensearch-and-sagemaker: A RAG-based Question Answering application using LLMs with Amazon SageMaker for model hosting and Amazon OpenSearch Service as the knowledge base.
rag-with-amazon-opensearch-serverless-and-sagemaker: A RAG-based Question Answering application using LLMs with Amazon SageMaker and Amazon OpenSearch Serverless as the knowledge base.
rag-with-amazon-postgresql-using-pgvector-and-sagemaker: A RAG-based Question Answering application using LLMs with Amazon SageMaker and Amazon Aurora PostgreSQL equipped with the pgvector extension for vector storage.
rag-with-knowledge-bases-for-amazon-bedrock-using-aurora-postgresql: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Knowledge Bases for Amazon Bedrock and Amazon Aurora PostgreSQL as the vector store.
rag-with-knowledge-bases-for-amazon-bedrock-using-L1-cdk-constructs: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Knowledge Bases for Amazon Bedrock created using L1 CDK constructs.
rag-with-knowledge-bases-for-amazon-bedrock: A RAG-based Question Answering application using LLMs powered by Amazon Bedrock, with Knowledge Bases for Amazon Bedrock as the knowledge base.
mcp-tutorial: A workshop for building and deploying Model Context Protocol (MCP) servers on local and AWS environments, and integrating them with AI applications. 👥

Data Analytics & Streaming (24)

amazon-emr-with-delta-lake: A guide to explore Delta Lake features on Amazon EMR, demonstrating how to read from and write to Delta tables.
aws-analytics-immersion-day: A workshop to implement a Business Intelligence (BI) system using various AWS Analytics Services like Kinesis, S3, Athena, OpenSearch, and QuickSight.
aws-athena-cqrs-pattern: An implementation of the CQRS (Command and Query Responsibility Segregation) pattern using Amazon Athena, where query execution results are delivered via email.
aws-dms-cdc-data-pipeline: Implements an end-to-end data pipeline to replicate transactional data from MySQL to Amazon OpenSearch Service via Kinesis using AWS Database Migration Service (DMS).
aws-dms-serverless-mysql-to-s3-migration: A data pipeline project that uses AWS DMS Serverless to migrate data from an Aurora MySQL database to Amazon S3.
aws-dms-serverless-to-kinesis-data-pipeline: A data pipeline project that uses AWS DMS Serverless to stream data from an Aurora MySQL database to Amazon Kinesis Data Streams.
aws-glue-streaming-etl-with-apache-iceberg: Creates a streaming ETL job in AWS Glue to build an updatable data lake on Amazon S3 using the Apache Iceberg format.
aws-glue-streaming-etl-with-delta-lake: Creates a streaming ETL job in AWS Glue to build an updatable data lake on Amazon S3 using the Delta Lake format.
aws-glue-streaming-ingestion-from-kafka-to-apache-iceberg: A collection of projects demonstrating how to ingest streaming data from Amazon MSK (Managed and Serverless) into Apache Iceberg tables on S3 using AWS Glue Streaming.
aws-msk-cdc-data-pipeline-with-debezium: Implements a Change Data Capture (CDC) pipeline to replicate data from MySQL to S3 using Amazon MSK, MSK Connect, and Debezium.
aws-msk-serverless-cdc-data-pipeline-with-debezium: Implements a CDC pipeline using Amazon MSK Serverless, MSK Connect, and Debezium to replicate data from MySQL to S3.
aws-opensearch-ingestion-tutorials: A collection of example projects for Amazon OpenSearch Ingestion, showing data ingestion into both OpenSearch domains and serverless collections.
opensearch-serverless-common-usage-patterns: A set of example projects for common Amazon OpenSearch Serverless use cases, including search, time-series analysis, and vector search.
redshift-streaming-ingestion-patterns: A collection of CDK projects demonstrating how to load data from streaming services like Kinesis and MSK into Amazon Redshift.
streaming-count-sketches-with-hyperloglog-in-amazon-memorydb: A project demonstrating how to efficiently count unique items in a data stream using the HyperLogLog algorithm in Amazon MemoryDB for Redis.
streaming-data-pipeline-from-kafka-to-s3-using-aws-kinesis-firehose: Example projects for managed data delivery from Amazon MSK to Amazon S3 using Amazon Kinesis Data Firehose.
transactional-datalake-using-amazon-datafirehose-iceberg: Implements a transactional data lake by ingesting CDC data from MySQL to S3 in Apache Iceberg format, using AWS DMS and Amazon Kinesis Data Firehose.
transactional-datalake-using-amazon-msk-and-apache-iceberg-on-aws-glue: Implements a transactional data lake by ingesting CDC data from MySQL to S3 in Apache Iceberg format, using Amazon MSK Connect (with Debezium) and AWS Glue Streaming.
transactional-datalake-using-amazon-msk-serverless-and-apache-iceberg-on-aws-glue: Implements a transactional data lake using Amazon MSK Serverless, MSK Connect (with Debezium), and AWS Glue Streaming to ingest CDC data from MySQL to S3 in Apache Iceberg format.
transactional-datalake-using-apache-iceberg-on-aws-glue: Implements a transactional data lake by ingesting CDC data from MySQL to S3 in Apache Iceberg format, using AWS DMS and AWS Glue Streaming.
web-analytics-on-aws: A simple web analytics system implemented with Amazon API Gateway, Kinesis Data Streams, Kinesis Data Firehose, S3, and Athena.
zero-etl-architecture-patterns: A collection of CDK scripts for creating Amazon RDS zero-ETL integrations with Amazon Redshift Serverless to enable near real-time analytics.
- aurora-mysql-to-redshift
- aurora-postgresql-to-redshift
- rds-mysql-to-redshift

Machine Learning & MLOps (17)

llava-next-video-model-on-sagemaker-endpoint: A CDK project to host the LLaVA-NeXT-Video model on an Amazon SageMaker endpoint for real-time and asynchronous inference.
llava-on-aws-sagemaker: A CDK project to host the LLaVA model on an Amazon SageMaker real-time inference endpoint.
video-llava-on-aws-sagemaker: A CDK project to host the Video-LLaVA model on an Amazon SageMaker real-time inference endpoint.
mlflow-ec2-sagemaker: A CDK project to provision an MLflow server on an Amazon EC2 instance.
mlflow-ecs-sagemaker: A CDK project to provision an MLflow server on Amazon ECS with AWS Fargate.
whisper-model-hosting-on-sagemaker-endpoint: Provides CDK projects to host the OpenAI Whisper model on Amazon SageMaker for real-time and asynchronous inference.
sagemaker-inference-component: A CDK project to deploy multiple foundation models to the same SageMaker endpoint using inference components.
deepseek-on-sagemaker: Provides a set of CDK projects to host DeepSeek models on Amazon SageMaker.
janus-pro-7b: A CDK project to host the Janus-Pro-7B model on an Amazon SageMaker real-time inference endpoint.
lgai-exaone-on-sagemaker/exaone-deep-7_8b-sglang: A CDK project to host LG's EXAONE model on Amazon SageMaker.
qwen-vl-on-sagemaker-endpoint/qwen2_5-vl-32b-sglang: A CDK project to host the Qwen-VL model on an Amazon SageMaker endpoint.
scale-to-zero-sagemaker-endpoint: Demonstrates how to deploy a SageMaker endpoint that can scale down to zero instances to reduce costs.
aws-realtime-image-analysis: A project for automatic image tagging and visualization using Amazon Rekognition, Lambda, Kinesis, and OpenSearch Service.
social-graph-based-people-recommender-using-amazon-neptune-and-textract: A people recommendation service based on a social graph built with Amazon Neptune, using Amazon Textract to perform OCR on business cards to build the graph.

SaaS & Architecture (1)

saas-metering-system-on-aws: A demo project for a SaaS metering system that enables usage-based billing models by accurately tracking API usage.

AWS CDK (1)

my-aws-cdk-examples: A comprehensive repository of 131 AWS CDK Python examples with architecture diagrams for a wide range of frequently used AWS services.

👥: Co-authored

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment