MindMemos
01AI-powered mental support journaling platform with real-time peer interaction, Socket.IO chat, and low-latency matching on a serverless AWS stack.
Node.jsAWS LambdaDynamoDBSocket.IO
About
Engineering with depth
I'm a recent grad from WashU, with hands-on experience across full-stack development, cloud infrastructure, and ML systems. I care most about scalable backends and real-time systems, the kind of work where architecture, performance, and reliability meet.
Skills
Tools & stack
Languages
- Java
- Python
- C++
- TypeScript
- JavaScript
Backend
- Node.js
- Express
- Spring Boot
Frontend
- React
- Next.js
- Angular
Databases
- PostgreSQL
- MongoDB
- DynamoDB
Cloud
- AWS
- Azure
- GCP
DevOps
- Docker
- Kubernetes
- CI/CD
ML
- PyTorch
- TensorFlow
Selected work
Flagship builds
MindMemos, MicroTutor, and Neighborly — full-stack and real-time systems. Thumbnails load from public/projects/ (see data/projects.ts).
MicroTutor
02Real-time micro-tutoring platform with WebRTC video, Socket.IO messaging, JWT auth, PostgreSQL, and Kubernetes-ready Docker deployment.
ReactNode.jsPostgreSQLWebRTCDockerKubernetes
Neighborly
03Location-based help platform with a map-driven Next.js frontend and a Spring Boot backend for discovering and offering neighborhood assistance.
Next.jsSpring BootTypeScriptMaps API
Experience
Where I've shipped
Work experience
Software Engineer Intern
Ignite Performance
- Full-stack athletic performance platform with an optimized backend and scalable AWS EC2–based infrastructure.
- Shipped production features with emphasis on reliability, API performance, and iterative delivery with the product team.
Teaching Assistant
Washington University in St. Louis
- Supported 50+ students weekly in Java, data structures, and debugging; improved assignment clarity through rubrics and structured office hours.
- Helped students resolve 100+ logic and memory issues, contributing to stronger completion rates and faster feedback loops.
Technology Intern
Aerostars UAV Technologies
- Improved full-stack website and Node.js/Express APIs, boosting engagement and streamlining onboarding for 100+ new users.
- Reduced load times through API and frontend optimizations and clearer data flow between services.
Achievements
Hackathons & competitions
HackWashU
PerfectBuy
First Runner-Up
AI-powered Chrome extension (MV3) predicting the best time to buy premium items — LLaMA 3.1 / Ollama with a privacy-first, rules-assisted engine; ~93% accuracy on expected price and wait-time signals in evaluation.
Stack: TypeScript, Chrome MV3, LLaMA 3.1, OPFS, Node.js
Research
Geospatial & ML
Satellite imagery and efficient deep learning for change detection.
NASA ADAPT–adjacent
GeoCompress
Geospatial change detection — NASA ADAPT–adjacent research
End-to-end satellite change-detection pipeline on OSCD, LEVIR, and DSIFN benchmarks. U-Net baseline reaching roughly 92% accuracy and 86% F1 on OSCD, compared against ChangeFormer-style architectures.
- Reproducible GitHub-driven workflow: preprocessing, batch inference, Acc/IoU/F1 reporting, and visualization across datasets.
- Exploring pruning and quantization (PTQ/QAT) targeting meaningful model size reduction while preserving near-baseline accuracy for edge (Apple M-series MPS) and GPU deployment.
PythonPyTorchTorchGeoJupyterOpenCVNumPy
Contact
Let's build something solid
Open to roles and collaborations in backend, full-stack, and real-time systems.