YW.Find a focus

Computer science · Research · Building with intent

Yun Waddy
Oo.

Computer Science student at Cal Poly working across AI/ML, information retrieval, and software engineering.

View projects Contact
Cal Poly · Class of May 2027Research and engineering
01 / Research
August 2026 - Present

Efficient Neural Retrieval

Undergraduate Researcher · Cal Poly CENG

Designing a reproducible evaluation pipeline to compare retrieval effectiveness, index memory, and query latency across dense-retrieval configurations.

Advisor · Maria Pantoja

Neural RetrievalRecall@knDCG@kRetrieval EvaluationEfficiency
Hover or focus to explore

What I did

  • Designed a reproducible evaluation pipeline for dense-retrieval configurations.
  • Used Recall@k and nDCG@k to evaluate ranking quality.

What I learned

  • Studied when efficiency choices create meaningful ranking-quality losses.

Impact

  • Centered the evaluation on retrieval effectiveness, latency, and index memory.
April 2026 - Present

Semantic Recommendations / Session Scout

Researcher & Technical Lead · Cal Poly CENG

Evaluated embedding models for personalized conference recommendation across more than 1,000 presentation titles and abstracts.

Advisor · Puneet Agarwal

EmbeddingsSemantic RetrievalRecommendation SystemsNext.jsFastAPIPostgreSQLAWS
Hover or focus to explore

What I did

  • Compared six embedding models using semantic ranking evaluation.
  • Used LLM-as-a-judge evaluation and user feedback.
  • Led the refactor from Streamlit to Next.js, FastAPI, PostgreSQL, and AWS.

What I learned

  • Evaluated how embedding choices affect personalized conference recommendations.

Impact

  • BGE-small-en-v1.5 achieved a mean relevance score of 2.4714.
  • Average recommendation retrieval time was approximately 0.07 seconds.
  • 95% of user-evaluation participants found the system useful.
October 2025 - March 2026

Belonging Beyond Boundaries

Machine Learning Research Assistant · Cal Poly CENG

Built Python data pipelines to preprocess, analyze, and visualize more than 2,600 EEG datasets.

Advisor · Sumona Mukhopadhyay

PythonEEGMNESignal ProcessingData Pipelines
Hover or focus to explore

What I did

  • Built Python data pipelines for more than 2,600 EEG datasets.
  • Implemented filtering, smoothing, and anomaly-detection techniques.

What I learned

  • Used MNE and related extraction tools to work with EEG data.

Impact

  • Focused the pipeline on reducing noise and artifacts for downstream analysis.
02 / Skills
01

AI / Machine Learning

TensorFlowscikit-learnHugging FaceLangChainGemini APIOpenAI APINLPEmbeddingsRecommendation SystemsClassificationNeural Retrieval
02

Data & Retrieval

pandasNumPyMNEPostgreSQLSQLVector SimilarityRetrieval EvaluationRecall@knDCG@k
03

Software Engineering

ReactNext.jsTypeScriptFastAPISpring BootExpress.jsREST APIsMongoDBMySQLMicroservicesSocket.IO
04

Cloud & Infrastructure

AWSDockerAzureRabbitMQSpring Cloud ConfigGit
05

Languages

PythonJavaCC++SQLTypeScriptJavaScript
03 / Work experience
January 2026 - Present

Web Developer

California Polytechnic State University - AIP

Developed responsive, WCAG-aligned web interfaces and worked on accessibility-focused software for Cal Poly.

  • Used semantic HTML, structured heading hierarchies, descriptive alt text, and accessible color contrast.
  • Partnered with DxHub AWS on an AI-powered executive dashboard.
  • The dashboard monitors accessibility compliance across all Cal Poly colleges and more than 4,000 courses.
AccessibilityWCAGWeb DevelopmentAWS
October 2025 - May 2026

Software Group Lead

Abbott x Society of Women Engineers

Led embedded-systems development for an Arduino-based medical-device clip-closing mechanism.

  • Worked on microcontroller software.
  • Supported hardware and software integration.
  • Focused on more consistent actuation.
ArduinoEmbedded SystemsHardware Integration
04 / Projects

Explore

Browse my work

Ideas, made real.

Projects grouped by the kind of work they represent.

AI, ML & Retrieval

Semantic Book Recommender

Built a semantic recommendation engine using embeddings and cosine similarity.

  • Reduced query latency by 60% through optimized vector storage.
  • Added zero-shot text classification for book categories.
  • Achieved 78% classification accuracy.
PythonLangChainGradioOpenAIpandasNumPyMatplotlib

Malicious URL Analyzer

Built a phishing-detection prototype combining rule-based URL analysis with machine-learning models.

  • Combined Logistic Regression and Random Forest.
  • Generated interpretable risk scores.
  • Detected more than 10 phishing indicators.
PythonStreamlitLogistic RegressionRandom ForestMachine Learning

Stock Price Predictor

Built an LSTM model for stock-price prediction using historical market data.

  • Performed exploratory analysis on data from nine companies.
  • Used a sliding 60-day input window.
  • Built a five-layer LSTM.
  • Used dropout and Adam optimization to reduce overfitting.
PythonTensorFlowKeraspandasNumPyscikit-learnMatplotlib

Credit Card Fraud Detector

Built a Random Forest fraud detector for an imbalanced transaction dataset.

  • Achieved 97.47% precision.
  • Achieved 87.01% F1 score.
  • Evaluated false positives and false negatives with a confusion matrix.
PythonpandasNumPyscikit-learnMatplotlib

Software Engineering

FitnessIQ

In Development

Building a full-stack fitness platform for workout tracking and nutrition planning.

  • Designed a microservices backend with four core services.
  • Uses Spring Boot and MySQL.
  • Uses RabbitMQ for asynchronous communication.
  • Uses Spring Cloud Config for centralized configuration.
  • Integrates Gemini API for AI-assisted recommendations.
JavaSpring BootReactMySQLGemini APIREST APIsDockerAWSRabbitMQMicroservices

Markr

Built a full-stack notes application supporting images, labels, and user-scoped data.

  • Owned the note-taking workflow across frontend, API, and database.
  • Implemented collaborative note editing for authenticated users.
ReactExpress.jsMongoDBREST APIsAzure

Real-time Chat Rooms

Built a real-time chat application using Socket.IO and bidirectional communication.

  • Messages update without page reloads.
  • Implemented backend routing with Express.
JavaScriptNode.jsExpressSocket.IOHTMLCSS

Open Source & Systems

Daft Open Source Contribution

Contributed PySpark-compatible to_degrees and to_radians aliases to Daft's numeric expression API.

Python
05 / Activities & hackathons

Keep learning.
Keep showing up.

AWS AI Summer CampAccessibility VPAT Assessment

Prototyped an AI-powered VPAT assessment workflow using AWS Bedrock.

Cal Hacks 2025Gret Support
Gemini 3 HackathonFind My Spot
Kiro HacksKiro Hacks
EventGoogle I/O 2026

Professional development · Technical community

EventAI Symposium 2026

Professional development · Technical community

EventUCLA Leadership Conference 2024

Professional development · Technical community

06 / Presentation

Oral presentation accepted

Session Scout: An AI-Based Decision Support Tool for Discovering Relevant Conference Presentations

2026 INFORMS Annual Meeting · San Francisco, California

Puneet Agarwal · T. Erb · A. Hill · Yun Waddy Oo · In preparation

07 / Credentials

Certifications

Data Science and ML Fundamentals Certificate
PythonpandasUnsupervised LearningClusteringClassificationRegressionPrediction
AI / Machine Learning Certificate
PythonNLPLLMsLangChainPineconeSpeech Recognition
AWS AI Practitioner Certificate
Foundation Model OptimizationAI SecurityComplianceGovernanceGenerative AI Solutions

Education

B.S. Computer Science

California Polytechnic State University, San Luis Obispo

May 2027 · GPA 3.8 / 4.0

Google Developer ProgramSociety of Women EngineersWomen in Software and HardwareCodePathHack Your Summer