// hi there, I'm

Khang Ho_

I'm a Software Engineer passionate about building full-stack applications and intelligent systems.

I enjoy turning complex problems into reliable, user-focused software while exploring the intersection of software engineering, machine learning, and deep learning.

View My Work Resume GitHub LinkedIn

01 / Where I've worked

Experience

Applied Research Center — FIU Engineering Center

Mar 2026 – Present

Undergraduate Research Assistant

  • Architected a vLLM-powered summarization pipeline that reduced technology article creation time from 5 days to 10 minutes.
  • Developed a FastAPI service using vLLM to convert Markdown summaries into structured JSON, reducing production failures by 15%.
  • Contribute to full-stack research applications including D&D KM and FAICE, supporting knowledge management and workflows for 40+ researchers.
PythonFastAPIvLLMASP.NETVisual BasicDockerLinux

State Farm

May 2026 – Aug 2026

Software Engineer Intern

  • Engineered an AI-powered CLI using Python, AWS Bedrock, and OpenSearch with a two-stage RAG pipeline, improving retrieval precision by 25% with <500ms latency.
  • Migrated enterprise CA certificates using Terraform and AWS Lambda, supporting infrastructure for 96M+ policies while maintaining zero downtime.
  • Developed a Kentucky-specific auto insurance coverage feature using Angular and TypeScript for 730K+ active auto policies.
PythonJavaScriptAngularTypeScriptAWSAWS BedrockOpenSearchAWS LambdaAWS CloudWatchTerraformCA CertificatesRAG

State Farm

May 2025 – Aug 2025

Software Engineer Intern

  • Led initial development of a multi-agent testing application using the OpenAI API, RAG, and vector databases, automating 80% of UI testing and reducing manual QA time by 30 minutes.
  • Optimized GitLab CI/CD pipelines by parallelizing automated test suites across policy coverage types (auto, fire, home), reducing deployment delays by 10 minutes.
  • Enhanced REST APIs with AWS Lambda fallback strategies and refactored 30+ unit tests, improving reliability for services supporting 1M+ policies.
JavaScriptNode.jsAWS LambdaAWS IAMAWS CloudFrontAWS CloudWatchTerraformOpenTofuREST APIsOpenAI APILLMsCI/CDBrunoAgile

02 / What I've built

Featured Projects

01

Football AI Detection

Computer Vision / AI · Feb 2026 – May 2026

A computer vision pipeline that detects and tracks soccer players, identifies teams, and projects player positions onto a tactical radar map.

PythonYOLOSigLIPByteTrackPyTorchOpenCVUMAPK-Means

Click to see engineering details

01 — Details

Problem

  • Coaches and analysts still rely on manually reviewing hours of match footage, where subtle details like positioning, spacing, and team shape are easy to miss. Those small details can be the difference between winning and losing.

Solution

  • An automated computer vision pipeline that processes raw match video end to end: detecting and tracking every player, classifying teams, and projecting positions onto a 2D tactical radar. Hours of manual film review become instant, objective spatial analysis.

Engineering Highlights

  • Player and object detection using YOLO
  • Team classification by clustering SigLIP visual embeddings with UMAP + K-Means
  • Object tracking using ByteTrack
  • Pitch keypoint detection
  • Homography-based coordinate transformation
  • Tactical radar visualization

Architecture

Input Soccer Video Frame / Video Processing Object Detection YOLO Player Cropping SigLIP Embeddings UMAP / K-Means Team Classification Tracking ByteTrack Pitch Analysis Keypoint Detection Homography Player Position Mapping Tactical Radar
View on GitHub ↗

Click to flip back

02

VPay

Full-Stack / AI / Voice Authentication · Sep 2025 – Nov 2025

A multi-agent payment platform that makes your voice your sole credential. It captures unique vocal characteristics to perform real-time identity verification, securely orchestrating hands-free payment processing without the need for physical devices.

PythonFastAPIGoogle ADKGemini 2.5 FlashReactTailwind CSSSQLite3Stripe API

Click to see engineering details

02 — Details

Problem

  • Paying today depends on carrying something: a phone, a card, a wallet. Forget them at home and you can find yourself in a critical situation with no way to pay, even though your identity never left you.

Solution

  • VPay makes your voice the credential. It converts speech into a multi-dimensional embedding unique to each user and verifies identity through embedding similarity before authorizing the transaction, so you can authenticate and pay with nothing but your voice.

Engineering Highlights

  • Voice embedding pipeline capturing unique vocal characteristics
  • Agent-driven authentication via embedding similarity
  • Multi-agent orchestration with Google ADK and Gemini 2.5 Flash
  • Full-stack build: FastAPI backend, React frontend, Stripe payments

Architecture

User Voice Speech Recognition Voice Embedding Embedding Comparison Agent Authentication (Google ADK) Payment Process (Stripe API)
View on GitHub ↗

Click to flip back

03

Cardvantage

Browser Extension / Full-Stack · Jul 2025 – Aug 2025

An intelligent Chrome extension that recommends the best credit card to use for an online purchase based on rewards, benefits, and user-specific factors.

Chrome ExtensionJavaScriptHTMLCSSGoogle GeminiGoogle OAuth

Click to see engineering details

03 — Details

Problem

  • Credit card reward programs contain complex rules across categories, multipliers, benefits, and loyalty programs, making it difficult to determine the optimal card for a purchase.

Solution

  • The extension captures relevant merchant information at checkout and sends it to a scoring API that evaluates available cards and returns a recommendation.

Engineering Highlights

  • Browser extension development
  • Checkout context extraction
  • API communication & recommendation logic
  • Reward optimization
  • User-focused product design

Architecture

Online Checkout Chrome Extension Merchant Information Extraction Google OAuth (User Sign-In) User Card Portfolio Gemini Card Evaluation Best Card Recommendation
View on GitHub ↗

Click to flip back

04

GhostB

Full-Stack / Machine Learning · Feb 2025 – Mar 2025

A budgeting app that targets psychology, not just numbers. Machine learning classifies each expense, and unnecessary purchases are inflated by up to 60%, recreating the friction of spending real cash.

PythonFastAPISQLAlchemyscikit-learnJavaScriptHTMLCSS

Click to see engineering details

04 — Details

Problem

  • Digital money doesn't hurt to spend. Balances are just numbers on a screen, so spending lacks the psychological friction of handing over cash, and by the end of the month it's hard to say where the paycheck went.

Solution

  • GhostB brings that awareness back. When a user logs an expense, a Random Forest classifier, pre-trained and then tuned on that user's own transaction history, labels it as needed or unnecessary. Unnecessary charges are inflated by up to 60%, so careless spending is felt immediately, and the inflated difference is banked as a ghost budget the user can claim back or allocate to savings.

Engineering Highlights

  • Full-stack financial management web application
  • User-specific training layered on a pre-trained Random Forest classifier
  • Real-time expense classification: needed vs. unnecessary
  • Cognitive spending mechanism inflating unnecessary charges up to 60%
  • Ghost budget system for claiming back or saving the difference
  • Backend APIs and database models for financial data

Architecture

Pre-Trained Random Forest Classifier User Transaction History User-Specific Model Training Needed / Unnecessary Classification Cognitive Charge Inflation (up to 60%) Ghost Budget & Savings
View on GitHub ↗

Click to flip back

03 / My toolbox

Engineering

Languages

PythonJavaJavaScriptTypeScriptSQLHTML/CSS

Full-Stack Development

FastAPIReactAngularNode.jsExpress.js

AI / Machine Learning

PyTorchscikit-learnRAGVector DatabasesvLLMOpenAI Agents SDKGoogle ADK

Cloud

AWS LambdaAWS BedrockAWS IAMAWS CloudWatchTerraformOpenTofu

Developer Tools

GitGitHubGitLabDockerLinuxCI/CDGoogle Colab

AI Developer Tools

Claude CodeCodexGitHub Copilot

04 / Giving back

Leadership

INIT FIU

Dec 2025 – Present

Explore Tech Lead — AI/ML

  • Design and lead hands-on AI/ML workshops using Google Colab and Notion.
  • Break down complex AI concepts into beginner-friendly modules.
  • Guide attendees through practical implementations.
10Workshops
20+Students / Session

INIT FIU

Sep 2025 – Dec 2025

Machine Learning Lead

  • Led a 5-person ML team within an 11-person cross-functional team building Prenergyze, an electric demand forecasting application.
  • Engineered and optimized ML pipelines processing 10K+ grid and weather data points.
18%Prediction Accuracy ↑
25%Preprocessing Time ↓
10K+Data Points

INIT FIU

Feb 2025 – May 2025

Web Development Member / Backend Lead

  • Led a 4-person backend team within a 14-person development team building DuelDot, a real-time multiplayer game.
  • Implemented map expansion, camera tracking, and real-time leaderboard functionality.
JavaScriptHTMLCSSNode.jsExpress.jsWebSocketSocket.io

05 / Selected for

Programs & Awards

Mastercard Mentoring Program — Cohort 6

Nov 2025 – May 2026

Selected from a competitive pool of applicants for a mentorship program featuring one-on-one mentorship with Mastercard executives.

JPMorgan Chase Career.edYOU Academy

Sep 2025 – Nov 2025

Selected from a competitive national applicant pool for an 8-week career development program focused on enterprise software engineering and fintech.

Jobright Campus Ambassador

Aug 2025 – Dec 2025

Selected from 1,000+ applicants to represent Jobright on campus.

200K+ Impressions · 90K+ Members Reached

Capital One Tech Summit

Jul 2025 – Aug 2025

Selected as 1 of 40 students nationwide for a 5-day engineering program focused on machine learning, web development, and product design.

Florida Medallion Scholars

FMS

Florida Medallion Scholars (FMS) award recipient.

06 / Where I study

Education

Florida International University

B.S. Computer Science · Minor in Artificial Intelligence

December 2027Expected Graduation
3.92GPA
Deep LearningMachine LearningData StructuresObject Oriented ProgrammingAlgorithm TechniquesDatabase ManagementNet-Centric ComputingOperating SystemsSystems ProgrammingComputer Architecture

07 / A little about me

About

I'm a Computer Science student at FIU, minoring in Artificial Intelligence, and I love building software that bridges traditional engineering and intelligent systems. I enjoy working across the stack, experimenting with new technologies, and turning ideas into useful applications.

Outside of building software, I enjoy leading technical workshops and helping other students get started with AI and software engineering.

Away from the keyboard, you'll find me on the badminton court, in the pool, or at the gym. Occasionally I'm back at the keyboard for a few rounds of CS2.

Currently Exploring

AI AgentsLLM SystemsComputer VisionFull-Stack Development

08 / Get in touch

Let's connect.

My inbox is open to everyone. Whether you have an opportunity to share, a question about my work, an idea to collaborate on, or just want to say hi, I'd love to hear from you.

KhangXHo@gmail.com
Resume GitHub LinkedIn