Incident root-cause analysis
I restored service reliability when production behavior diverged from healthy-looking checks. I traced logs, deployment state and request paths, verified the fix, and documented where my first diagnosis was wrong.
Seunghyun Ryu · 류승현
Site Reliability Engineer for regulated industries
AI-assisted operations for cloud reliability and cloud security on GCP, from incident response to FinOps. My experience spans regulated industries: public sector, finance and medical devices. I use evidence to guide changes.
Selected work
Operations patterns from regulated industries
Cloud security governance
I brought cloud security controls into one reviewable pipeline in a regulated environment. I connected asset inventory, logging, risk triage and evidence reporting to give ongoing security work a clear trail.
Cloud migrations
I brought services from separate hosting environments into GCP. I planned DNS and traffic changes, checked service behavior and prepared rollback paths, bringing infrastructure under a shared operating model.
Cloud cost management (FinOps)
I reduced cloud waste without treating every billing drop as a saving. I compared billing with sustained usage, adjusted resource sizes and automated reports that separated real savings from currency effects and billing gaps.
AI ChatOps and model observability
I made cloud queries easier through AI ChatOps: conversational operations tools. I added read-only query boundaries and measured large language model (LLM) latency, failures and token use to guide changes rather than guess.
How I work
Evidence, approval and clear boundaries
Evidence before conclusions
I compare logs, deployment state and observed behavior before calling a problem resolved. If the evidence changes, I revise the diagnosis.
People approve lasting changes
I keep human approval in the path for irreversible changes. Automation can prepare a change; a person decides whether to apply it.
Clear data boundaries
I explain methods without exposing private systems or data. Public examples use public or synthetic material, with their limits stated.
Experience
Cloud operations across regulated sectors
SRE means Site Reliability Engineering.
Jul 2025 – present
Cloud Engineer (SRE · Cloud Security)
Medical-device company
Apr 2023 – Jun 2025
Site Reliability Engineer
Cloud managed-service provider (banking and public-sector clients)
May 2022 – Apr 2023
Site Reliability Engineer
Cloud managed-service provider (predecessor in the same group)
Jan 2022 – May 2022
Site Reliability Engineer
Cloud managed-service provider (public-sector clients)
Certifications
Earned CKA (2022) and CKAD (2023).
Toolbox
Google Cloud Platform (GCP) · Kubernetes · Cloud security
Terraform · Infrastructure as Code (IaC) · GitOps · Argo CD · CI/CD
Linux · Docker · Python
Earlier experiments
Small public projects exploring source-backed answers, evaluation and local prompt analysis.
cite-or-refuse
A documentation question-answering experiment that cites a source or declines to answer.
ProvenanceBench
A benchmark exploring faithful answers and when declining to answer is the right result.
promptprint
A local tool for reviewing how you prompt AI coding agents and what to improve next.