About

Sieon Lee · AI Engineer · Data Scientist

I build the retrieval, pipeline, and agent infrastructure that keeps AI systems honest in production.

I’m an AI engineer and applied data scientist based in Chicago. Most of what I write about on Sieon Labs comes from systems I’ve shipped or am actively running, not from summarizing someone else’s paper. My focus is the layer between a model and a product: retrieval quality, pipeline reliability, evaluation, and the agent scaffolding that turns an API call into something dependable.

AI agents Retrieval & RAG ML infrastructure Data engineering Healthcare AI
Now Building a healthcare RAG capstone at UChicago and shipping agent workflows at Super Amplify.
3+ yrsProduction data and ML systems, not tutorial projects.
500GB+Vector index in production on AWS OpenSearch k-NN.
UChicagoM.S. Applied Data Science — ML, NLP, Bayesian modeling, cloud, healthcare AI.

What I actually work on

Model quality is rarely the bottleneck in production AI. Retrieval quality, pipeline reliability, evaluation harnesses, and observability usually are, and that’s the layer I spend most of my time in. I’ve owned that stack end to end: ingestion, embedding, indexing, query-side tuning, and the monitoring that tells you when any of it silently degrades.

At the University of Chicago, my capstone is a healthcare AI retrieval system: structured clinical data joined against unstructured clinical notes, indexed for hybrid dense and sparse retrieval, with a RAG layer for question answering evaluated against held-out clinical QA data.

Systems I’ve built

Hermes

My personal AI operating system: an agent runtime wired to tools, skills, persistent memory, and a retrieval layer over a personal knowledge base. The sync design treats one store as the single source of truth and keeps local indexes as disposable, rebuildable projections rather than a second copy of truth. I use it daily for research, writing, and publishing, this page included.

Vector search at scale

A 500GB+ embedding store on AWS OpenSearch k-NN serving similarity search for production workloads at Nexon Korea. Owned the pipeline end to end, from ingestion and embedding generation through index lifecycle management and query-side tuning.

Data & cloud infrastructure

Python ETL and SQL Server auditing pipelines, Terraform-managed cloud databases, and Grafana-based monitoring, built for teams that need pipelines to fail loudly and recover cleanly instead of dropping rows silently.

Healthcare RAG (capstone)

Structured and unstructured clinical data behind a hybrid retrieval and RAG question-answering system, built and evaluated as part of my M.S. capstone at the University of Chicago.

Experience

Super Amplify — AI & Agent Development Intern

Building integrations, execution workflows, intent-routing logic, and retrieval architecture for an enterprise AI agent platform.

Nexon Korea — Data Engineer

Designed production-scale data and ML infrastructure: Python ETL, SQL Server auditing, Terraform-managed cloud databases, Grafana monitoring, and a 500GB+ vector embedding store on AWS OpenSearch k-NN.

LG CNS Europe — IT MES Systems Engineer

Optimized SQL procedures for real-time manufacturing dashboards, built C# automation tools, and improved operational reporting workflows.

Stack

AI & Data Science
PythonRPyTorch TensorFlowscikit-learnPandas NumPyNLPBayesian modeling
Data Systems
SQLPySparkETL/ELT Batch & streamingData modelingData quality monitoring
Cloud & Infrastructure
AWSAzureGCP BigQuery OpenSearchAurora MySQLRedis TerraformDockerKubernetesGrafana

Why Sieon Labs exists

This is where I document practical AI engineering, not model announcements or API demos, but how AI systems are actually wired together: agents, retrieval, data infrastructure, ML pipelines, cloud systems, and the operational details that make AI products reliable.

If a post here doesn’t map to something running somewhere, it doesn’t get published.

Contact

If you’re interested in AI agents, applied data science, retrieval systems, or the infrastructure behind modern ML products, feel free to reach out.