// software engineer · C/C++ backends · cloud migration

I migrate platform backends to the cloud
and keep them fast and stable in production.

Yoomin Oh — Software Engineer

I move platform backends into AWS and make the C/C++ services behind them faster and simpler to run — migrations, legacy technical debt, and the secure APIs on top. Give me a system that's slow or stuck on aging infrastructure and I'll hand back one that's fast, well-instrumented, and boring to operate.

View work Contact me

~/about

Problems first, stacks second.

A short version of how I work and what I care about.

I'm a software engineer working on cloud migration and the performance of C/C++ backend services. Since 2023 I've worked on the in-flight entertainment platform at Panasonic Avionics — moving it into AWS (EKS to ECS, and off on-prem hardware), paying down technical debt in the C/C++ server, and building the secure REST API layer other applications use to talk to it.

I care about instrumentation before optimisation, automated verification over manual checking, and cutovers that move production traffic without downtime or data loss. I pick up unfamiliar systems quickly and I'm willing to do the tedious, unglamorous work that getting something into production actually takes. I hold the AWS Developer Associate certification.

years shipping production systems

~95%

data transmission success to aircraft, up from ~20%

EKS→ECS

plus on-prem to AWS migration, validated for zero-downtime cutover

AWS

Developer Associate certified

~/skills

Technical toolkit

Chosen per problem — this is what I reach for most often.

Languages & systems

C / C++Python GoShell LinuxSocket programming ConcurrencyMultithreading Legacy refactoring

Networking

TCP/IPHTTP/HTTPS TLS/SSLDNS REST APIsJWT auth Subnetting

AWS & infra

ECSEC2 ECRRDS IAMCFT S3AWS WAF LambdaLambda authorizers API GatewayRoute 53 VPCSecurity Groups Load balancingHorizontal scaling CloudWatchFireLens Docker

Delivery & tooling

CI/CD pipelinespytest GoogleTest (gTest)SonarQube Linux debugging toolsJIRA JAMA

~/projects

Selected work

Four efforts at Panasonic Avionics where the interesting part was the constraint, not the feature.

EKS → ECS migration validation

Problem: The in-flight entertainment platform ran on EKS with an under-provisioned cluster topology: worker nodes intermittently failed to join the control plane due to VPC/CNI networking faults, and CPU was contended across a shared ~20-node pool with no effective resource isolation, producing unpredictable latency. Together with the cost of the over-provisioned cluster, the instability was disrupting the airline customer's operations.
Solution: Migrated services from EKS to ECS, configuring per-task CPU/memory allocation for guaranteed resources (vs. shared-node contention) and FireLens log routing for near-real-time logging, and owned end-to-end validation testing to confirm functional and performance parity with the existing EKS setup before go-live.
Impact: Data transmission success to aircraft improved from ~20% to ~80% after cutover, with no disruption to clients already running in production — restoring a customer revenue stream that platform instability had undermined.

AWS ECSAWS EC2 AWS IAMAWS Security Group AWS VPCAWS Route 53 AWS AthenaAWS CloudWatch AWS FirelenAWS CFT CI/CDDocker

Proprietary — Panasonic Avionics

Scaling out a legacy C/C++ platform server

Problem: The C/C++ server behind the in-flight entertainment platform ran on a single instance and had built up years of technical debt — dead code, race conditions, deadlocks, and functional bugs. It couldn't scale out, and every change was risky.
Solution: The legacy code was brought under a gTest test suite, then its data races, deadlocks and functional bugs were fixed and dead code removed. The server was then split from one process into a load-balanced pool of instances.
Impact: Data transmission success to aircraft improved from ~80% to ~95% after the fixes, with a marked drop in field-reported issues. Eliminating the data races and deadlocks removed the stalls and dropped connections that had been failing transfers, while the three-instance pool absorbed peak load that previously saturated the single process. The gTest suite now gates changes in CI, so the same classes of defects are caught before release.

C/C++Linux MultithreadingConcurrency gTestCI/CD Load balancingHorizontal scaling Legacy refactoring

Proprietary — Panasonic Avionics

Secure Rest APIs on AWS

Problem: The platform server exposed only a low-level socket API, so every new application that needed to integrate had to be built against C/C++ socket code. This raised the barrier to onboarding new applications designed for the specific customer, tied integration timelines to scarce systems-programming expertise
Solution: Built the REST API on AWS with a Lambda backend inside a private VPC. Access is secured with JWT bearer tokens issued by Panasonic Corporation's identity provider; a custom Lambda authorizer decodes and validates each token, authorizes the caller, and caches results to keep latency low. Delivered through a CI/CD pipeline with pytest unit coverage and SonarQube static analysis as quality gates.
Impact: New applications integrate over standard authenticated REST with no C/C++ socket code, so customer-specific features reach production faster and integration no longer depends on low level c/c++ systems-programming expertise.

AWS API GatewayAWS Lambda AWS IAMAWS ACM AWS VPCIdentity Provider AWS Route 53 JWTCI/CD PythonPytest Unit TestSonarQube

Proprietary — Panasonic Avionics

On-prem → AWS migration of a legacy platform

Problem: The legacy platform ran entirely on a single on-premises server — the database and two software components all shared the same box. Starved of CPU and memory, the system was slow enough that a single API call could take up to an hour to return, far outside the customer's requirements and expectations.
Solution: The database and both components were migrated from the on-prem server to AWS and sized with enough CPU and memory headroom to run comfortably. Database contents were replicated with AWS DMS and file assets with AWS DataSync to keep the two environments in sync through the transition. Alongside the move, the application code was profiled and optimized to cut latency across all API calls. The cutover was planned and validated to move production traffic across without downtime or data loss.
Impact: Average API latency dropped from ~5 minutes to a few seconds once the workload had adequate CPU and memory and the query paths were optimized, with worst-case calls no longer running for an hour. Downstream platform applications can now query the API in near real time instead of blocking on long-running requests, and the cutover completed with no downtime and no data loss. The result was recognized by leadership and cited as a competitive advantage in customer conversations.

AWS ECSAWS EC2 AWS RDSAWS VPC AWS IAMAWS CloudWatch AWS DMSAWS DataSync Legacy migrationCutover planning Performance optimizationDocker CI/CDPHP

Proprietary — Panasonic Avionics

~/experience

Where I've worked

Recent roles and the milestones that mattered.

May 2024 — Present

Software Engineer

Panasonic Avionics Corporation · Irvine, CA

Cloud modernization and core-service performance work for the in-flight entertainment platform.

Re-platformed from EKS to ECS with per-task CPU/memory allocation and FireLens logging, and owned validation for a zero-downtime cutover; data transmission success to aircraft rose to around 95%.
Migrated the legacy stack off on-prem hardware into AWS, using AWS DMS and DataSync to move data and files, which brought average API latency down from minutes to seconds.
Cleared long-standing technical debt in the C/C++ server — race conditions, deadlocks and dead code — behind a new GoogleTest suite, then scaled it from a single instance to a load-balanced pool.
Built the REST API layer on API Gateway and Lambda inside private VPCs, secured with JWT and a custom Lambda authorizer and delivered through CI/CD with pytest and SonarQube quality gates.

Feb 2023 — Apr 2024

Software Test Engineer

Panasonic Avionics Corporation · Irvine, CA

Functional and integration testing for the in-flight entertainment platform.

Wrote and executed technical validation cases in JIRA and JAMA, keeping test coverage traceable to product requirements.
Replaced manual test workflows with automated validation scripts, shortening the feedback loop on each build.
Ran cross-platform hardware/software integration testing against functional specifications.

Dec 2022

B.S. Computer Science

University of California, San Diego · La Jolla, CA

Graduated December 2022.

~/contact

Let's talk

Hiring, contracting, or just want to compare production incident stories — send a note.

Email click to copy LinkedIn View profile →