I'm a computer science student at UVA Engineering with a strong interest in ML security, edge-to-cloud architectures, and building software tools that help people. Here's my personal website! https://elliotschmango.github.io/elliothong_aboutme/

• Machine Learning Sec (esp. LLMs & analytics)
• Cloud engineering (AWS/GCP)
• Mobile/Webdev development with React Native/Expo, React JS, Django, Flask
• Computing research in various fields

• Weightlifting & golf ⛳
• Motorsports & sim racing 🏎️
• Guitar
• Cooking

Waste Water Treatment Plant - Anomaly Detection System for water infrastructure monitoring

• Deploy-Everywhere Architecture: Refactored from a rigid Jetson prototype to a fully Dockerized stack capable of running on standard x86/ARM hardware (JetPack, Linux, Windows, etc).
• Hybrid AI Pipeline: Features a novel "Edge-to-Cloud" inference engine using a local SSIM-Denoising Autoencoder for real-time detection and Gemini 3 Pro for zero-shot anomaly classification.
• Modern Tech Stack: Replaced legacy PyQt5/I2C controls with a reactive Streamlit web dashboard and universal ONVIF/RTSP standards for commercial security cameras.
• Advanced Reliability: Implemented thread-safe state management and a rigorous "One-Shot" benchmarking framework for validating detection sensitivity in dynamic water environments.

Scalable Telemetry & Operations ML Platform is a secure, containerized microservices platform for anomaly detection and alerting in telemetry data. Inspired by mission-critical systems used in defense, utilities, and autonomous ops.

• Modular microservice architecture enables independent scaling and high availability
  • Built with FastAPI services for authentication (JWT) and strict role-based access control (RBAC)
  • Simulates real-time telemetry streams (process failures/custom metrics) with injected anomalies
  • Uses ML-based anomaly detection to flag abnormal patterns in operational data
  • Includes Prometheus monitoring, Grafana dashboards, email alerts, and centralized logging (Fluentd)
• Cloud-compatible with Dockerized services, AWS deployment, and load balancing support

A mock streaming platform built with React and Django, developed by a team using Agile and sprint-based workflows

• Simulates Netflix-style features including video browsing, playback, and user profiles
• Includes auth, watch history, favorites, and role-based admin controls
• Developing with a 6-sprint Agile process and clearly defined roles (Scrum Master, DevOps, Testing, Requirements)
• AWS used for cloud storage (S3) and deployment, with CI/CD pipeline integration

A mobile skin condition classifier app built with React Native + AWS + Express

• Uses AWS SageMaker-hosted ResNet18 model to diagnose skin conditions from photos
• React Native app built with Expo; supports camera & image uploads
• Uses AWS DynamoDB for APA-style sources and dynamic, inclusive UI for condition explanations

Technology lending system for a library

• Worked as the requirements manager in an agile development scrum-based team.
• Built with Django backend and Tailwind frontend
• Led requirements gathering and stakeholder interviews
• Implemented ticketing logic and borrowing workflows

Data science project analyzing population trends across Virginia counties

• Collected and cleaned multi-year demographic data from the Virginia Open Data Portal
• Built regression and time-series models (LSTM, polynomial regression) to forecast county-level population shifts
• Used PCA for dimensionality reduction and K-Means clustering for pattern discovery across regions
• Visualized trends and anomalies using Matplotlib and Seaborn to inform future infrastructure planning

Multimodal assistant for water plant engineers

• Built with React (Vite) frontend and Django backend, integrated via REST API
• Uses Llamafile’s Llama3.2-3B model for local inference with no external API calls for security
• Supports Retrieval-Augmented Generation (RAG) and iterative Co-RAG querying for improved contextual understanding
• Uses BLIP-2 for chart interpretation and AWS Amazon Textract for OCR
• Integrates with Microsft Azure for authentication/company security

Role-based HR dashboard to optimize personnel assignment across projects

• Tech Stack: React.js, Django, and MongoDB
• Designed and implemented a custom algorithm inspired by bipartite matching
• Matched staff to initiatives based on Level of Effort (LOE), hiring status, and project budgets
• Built dynamic dashboards with secure login, LOE tracking, and administrative controls

Finance-tech project for managing personal and business expenses

• Built with Next.js and React frontend for responsive user experience
• Integrated Plaid API for secure banking data aggregation
• Implemented Appwrite Authentication and Sentry for security and error monitoring
• Designed for extensibility with future transaction categorization and budgeting tools

A local document-based chatbot built with Streamlit, Ollama, and LangChain

• Lets users upload and query their own PDFs using a fully local RAG pipeline
• Uses LangChain to chunk and embed documents into a vector store for semantic search
• Runs a locally hosted LLM via Ollama to generate context-aware responses
• Prioritizes offline use and privacy with no cloud dependencies

• GCP Associate Cloud Engineer - In Progress
• AWS Certified Solutions Architect - Associate: In Progress
• AWS Certified Cloud Practitioner - In Progress
• PCEP Certified Entry-Level Python Programmer: 2023

Thanks for visiting my profile! Feel free to explore my repos or reach out.