YJ
Elevation · Three entries

Experience

A vertical record of professional roles — each chapter filed against the timeline that holds them together.

Chapter 03

ALTR

Software Engineer I

Go · Node.js · Java · AWS Lambda · Step Functions · DynamoDB · RDS · S3 · SQS · Terraform · Redis

  1. Engineered a greenfield data-governance platform as event-driven microservices on Lambda, SQS, and Step Functions — shipping v1 a week early and enforcing masking and access policies across customer OLTP databases in production.
  2. Designed a Go JWT authorizer conforming to RFC 7519 and 8725, paired with a Java client SDK, to secure every request flowing through the API Data Gateway.
  3. Cut costs by $10,000+ and query latency 90%+ by migrating hot reads to Redis, indexing RDS tables, and redesigning DynamoDB key layouts around real access patterns.
  4. Ported legacy Java ECS/gRPC services to serverless Lambda and Step Functions through Terraform, collapsing infrastructure complexity while preserving behavior.
  5. Built observability from scratch across every serverless pipeline — structured CloudWatch logs, metrics, and alarms wired to the operational surface engineers actually watch.
Chapter 02

Federated Learning of Medical Image Reconstruction

Undergraduate Researcher

Python · PyTorch · Flower

  1. Improved accuracy on a SPECT medical-imaging ML pipeline by generating synthetic training data demographically aligned with underrepresented patient cohorts.
  2. Automated data cataloging and augmentation so the model could train faster and on a broader slice of the population than the original dataset afforded.
  3. Integrated federated learning with Flower, enabling secure decentralized training across contributors with per-node authentication.
  4. Evaluated deployment strategies for secure model transport and UI access across the federated network of participating nodes.
Chapter 01

NEural TransmissionS Lab

Undergraduate Researcher

Python · PyTorch · Blender · R · MATLAB

  1. Integrated drone vision, navigation, and control into a single Python pipeline, enabling a swarm of drones to dock on pieces of space debris.
  2. Applied matrix projections to camera input, turning raw point clouds into artificial-potential-field vectors for collision-free rendezvous.
  3. Developed a space-debris simulation pairing the YOLOv9 object detector with 3D Gaussian Splatting to reconstruct target geometry.
  4. Validated real-time control and vision performance of the MARVIN autonomous system on an NVIDIA Jetson Orin.