I build things
I want to understand.
I’m someone who enjoys learning deeply, trying new things, and following interesting questions wherever they lead.
Visitor orderThe default pass keeps every section equally prominent. All content remains unchanged.
RS
VISITOR PASS
VisitorDEF-00 · PORTFOLIO 2026Drag the pass and let it swing.
01 / ABOUT
Overview
Hey y’all, I’m Ryan! I’m a computer science student at UT Austin interested in AI research, machine learning, and building tools that turn interesting ideas into something real. I’ve taken a pretty unconventional path through school, moving through an accelerated track and finishing two years ahead of my cohort, which gave me a lot of freedom to explore things on my own. I’m especially drawn to problems where I can dig into the research, experiment, and figure things out as I go. Outside of computer science, you’ll usually find me reading web novels, working out, or chasing new personal records in roguelikes.
Current education
B.S. in Computer Science
UT Austin
Expected year of graduation: 2029
Certificate in Quantum Information Science
Fields of Interest
Areas I return to across research, products, and problem-solving.
Entrepreneurship
Turning promising ideas into practical, scalable ventures.
Business Management
Strategy, operations, growth, and teams.
Artificial Intelligence
Learning systems, intelligent agents, and applied research.
Computer Science
Software systems, algorithms, and useful tools.
Mathematics
Proof, problem solving, and quantitative reasoning.
02 / AWARDS
Tap the front tile to advance
- 01USAMO Qualifier
12th grade · National
- 02AMC 12 Perfect Score
11th–12th grade · National
- 03AMC 10 Distinction
10th grade · National top 2.5%
- 04UIL Computer Science State Runner-Up
12th grade · Texas state competition
- 05Greater Austin Regional Science Fair · 3rd Place
11th grade · Physics & Astronomy
- 06ASCL Intermediate Division · International 1st Place Team
11th grade · International
- 07Log1 National Math Contest · 1st Place in Texas
11th grade · Statewide
- 08DECA ICDC Qualifier · Three-Time
10th–11th grade · Texas
- 09DECA ICDC Retail Merchandising Finalist
10th grade · International
- 10DECA ICDC Exam Top Performer
10th grade · International
- 11BEST Robotics State · 3rd Place
9th grade · Texas
- 12Texas HOSA State Qualifier
10th grade · Texas
- 13Jane Street Estimathon · Top 12
11th grade · Quantitative problem solving
03 / PROFESSIONAL EXPERIENCE
06/2025 — 03/2026
01
Software Engineering + Marketing Intern
Jammy Chat
Built production-oriented machine learning, recommendation, and retrieval systems alongside the services and evaluation infrastructure needed to operate them.
- Built an LLM-powered RAG system using embeddings and pgvector semantic search for context-grounded FAQ responses
- Developed PyTorch and scikit-learn NLP pipelines for sentiment, category, and urgency classification
- Created a personalized music recommendation engine using Spotify data, user embeddings, similarity search, and ML ranking
- Engineered FastAPI services with PostgreSQL, Redis, async processing, Spotify OAuth/API integration, and Docker
- Created ML and LLM evaluation pipelines for retrieval quality, classification, recommendation relevance, and inference latency
03/2025 — 09/2025
02
Technical Outreach Manager
Inclusive Computing Initiative
Led the organization’s technical outreach, brand strategy, digital campaigns, and expanding national chapter network.
- Rebranded ICI and led three outreach campaigns
- Organized Dallas’s first Inclusive Hackathon and secured more than $1,000 in corporate sponsorships
- Produced 60+ posts across five social media platforms
- Designed and organized a Facebook Ads management platform
- Trained and managed 10 interns while overseeing marketing for five national chapters
03/2025 — 06/2025
03
Growth Intern
Agrisaath
Selected as one of five interns from hundreds of applicants to collaborate with University of Pennsylvania students on an international agri-startup initiative.
- Published 30+ blog posts for Instagram and LinkedIn to expand the organization’s reach
- Designed corporate marketing and branding materials supporting outreach across 10 countries
- Developed brand positioning and copy for four international campaigns
- Built a content calendar containing 20+ reels, posts, and blogs for consistent engagement
04 / PROJECTS
ALL PROJECTS · DRAG TO EXPLORE
Move around the wall, then tap any tile to locate it in the searchable catalogue below.
21 results
01
AI + Retrieval
CLIP Photo Search
A private, local natural-language photo search engine with Python and Streamlit on desktop plus a React Native mobile client, using MobileCLIP, ONNX, SQLite, and vector retrieval.
- Python
- PyTorch
- TypeScript
- React Native
- Expo
- Streamlit
- SQLite
- ONNX
- Vector search
02
C++ + Systems
File-Backed KV Store
A C++20 command-line database with an append-only checksummed log, POSIX locking, CRC-32 recovery, atomic compaction, integration tests, sanitizers, and Ubuntu CI.
- C++20
- POSIX
- CMake
- Linux
- GitHub Actions
03
AI + Retrieval
Repo Recall
A C++20 hybrid retrieval engine that maps software incidents to historical fixes using BM25, MiniLM embeddings, HNSW, reciprocal-rank fusion, gRPC, and Python evaluation tooling.
- C++20
- Python
- ONNX
- Vector search
- Machine learning
- gRPC
04
C++ + Systems
Linux Job Runner
A C++20 Linux process supervisor built around fork, exec, pipes, resource limits, signals, restart policies, JSON summaries, sanitizers, and continuous integration.
- C++20
- POSIX
- Linux
- CMake
- GitHub Actions
05
Quant + Data
Earnings Call Analyzer
A point-in-time NLP research pipeline that tests 10-Q and 10-K language against the next eligible market session with temporal splits, classical models, and frozen FinBERT baselines.
- Python
- NLP
- PyTorch
- scikit-learn
- Machine learning
06
Quant + Data
Information Diffusion
A Python and C++20 market-microstructure research system for event-aligned studies of CPI information flow across Treasury and E-mini S&P 500 futures.
- Python
- C++20
- Machine learning
05 / PROJECT-BACKED TECH STACK
01 / LANGUAGES
02 / AI + RETRIEVAL
03 / FRONTEND + MOBILE
04 / BACKEND + DATA
05 / SYSTEMS + INFRASTRUCTURE
06 / PLATFORMS + RESEARCH
21 PROJECTS · 20 PUBLIC REPOSITORIESView GitHub profile ↗
Skills backed by shipped work.
Select any technology above to jump directly to the projects that use it. Public repository links are included where available; the Temple CRM remains private.
06 / RESEARCH
01COMPLETED · 2026 · PHYS-0786-SR-TCharacterizing Population-Level Trends in Type 2 AGN Activity
An observation-driven study of how local Type 2 active galactic nuclei group in measurable optical and infrared parameter space—and how those populations change across redshift.
+
Characterizing Population-Level Trends in Type 2 AGN Activity
An observation-driven study of how local Type 2 active galactic nuclei group in measurable optical and infrared parameter space—and how those populations change across redshift.
Abstract
Type 2 Active Galactic Nuclei (AGNs) are a key component for understanding black hole growth because dust obscuration can hide signatures that are easier to detect in unobscured AGNs, biasing many large-sample evolution studies. This project develops an observation-driven method to track how Type 2 AGN populations change over cosmic time using only directly measured quantities. We construct a clean sample of local Type 2 AGNs (z < 0.2) using optical spectroscopy from the Sloan Digital Sky Survey (SDSS DR16), cross-matched with mid-infrared photometry from the Wide-field Infrared Survey Explorer (WISE), and supplemented with public catalog information. From these data we compute emission-line ratios, [O III] luminosity, and WISE color indices, then apply unsupervised machine learning—Principal Component Analysis (PCA) and UMAP for dimensionality reduction, followed by K-Means clustering—to identify distinct groupings (“stages”) in observable parameter space. We compare cluster abundances and cluster-averaged properties across redshift bins and validate interpretations with BPT diagnostic diagrams to separate AGN activity from star-formation contamination. Our approach reveals multiple Type 2 AGN clusters with different power and dust/torus signatures and shows measurable shifts in cluster composition with redshift, indicating evolutionary trends even within the nearby universe. Scientifically, this provides astronomers a utilizable framework for discovering population evolution in survey-scale datasets and prioritizing targets for follow-up spectroscopy. More broadly, it demonstrates how machine learning can convert large catalogs into interpretable insight, supporting the public understanding of black hole growth and cosmic history.
Methods · SDSS DR16 optical spectroscopy cross-matched with WISE mid-infrared photometry; PCA and UMAP dimensionality reduction; K-Means clustering; BPT diagnostic validation.
Finding · Type 2 AGNs form structured subgroups with distinct power and dust/torus signatures, and cluster composition shifts measurably across nearby-universe redshift bins.
- SDSS DR16 + WISE
- PCA · UMAP · K-Means
- BPT diagnostics
- Python
02COMPLETED · 2025 · PHYS-0618-SR-TRedshift and the Properties of Type 2 Active Galactic Nuclei
A machine-learning analysis of 255 Type 2 AGNs examining how luminosity, color, and host-galaxy properties vary with redshift and cosmic history.
+
Redshift and the Properties of Type 2 Active Galactic Nuclei
A machine-learning analysis of 255 Type 2 AGNs examining how luminosity, color, and host-galaxy properties vary with redshift and cosmic history.
Abstract
The exploration of Active Galactic Nuclei (AGNs), the highly energetic centers of galaxies hosting supermassive black holes, has been a cornerstone of modern astrophysics. Milestones such as the discovery of quasars in the 1960s and the identification of the first AGN in a Seyfert galaxy have established AGNs as key tools for studying galaxy evolution and black hole growth. Despite decades of research, the mechanisms linking AGN properties to cosmic distances remain untouched. This project investigates the evolution of Type 2 AGNs across the redshift spectrum, where redshift serves as a critical measure of distance and cosmic history. We utilized data from NASA’s HEASARC, specifically the XMM-CTY2AGN database, comprising multi-wavelength spectroscopic observations of 255 Type 2 AGNs. Employing advanced machine learning models like Gaussian processes and x-y kernel methods, we analyzed non-linear relationships between redshift and AGN properties. Our findings reveal that Type 2 AGNs at higher redshifts exhibit significantly greater luminosities and elevated U-V color indices. These results suggest an evolutionary trend in AGN characteristics, shedding light on their role in shaping the early universe and advancing our understanding of supermassive black hole activity.
Methods · 255 XMM-CTY2AGN sources from NASA HEASARC; Gaussian Process Regression, Spearman analysis, LightGBM, Lasso, Random Forest, and an 80/20 train-test split.
Finding · Higher-redshift Type 2 AGNs showed greater X-ray luminosity and elevated U-V color indices; Gaussian Process Regression produced the strongest tested fit (R² ≈ 0.659).
- 255 Type 2 AGNs
- NASA HEASARC
- Gaussian processes
- Regression analysis
07 / PUBLICATIONS
BK / 01Competitive Finance VocabularyRYAN SHANKAR
Books
Competitive Finance Vocabulary: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackBK / 02Hospitality & Tourism VocabularyRYAN SHANKAR
Books
Hospitality & Tourism Vocabulary: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackBK / 03Business Admin. Core KPIsRYAN SHANKAR
Books
Business Admin. Core KPIs: Performance Metrics for Business Growth and Competition
Competitive Business KPIs · Kindle, hardcover & paperbackBK / 04Competitive Entrepreneurship VocabularyRYAN SHANKAR
Books
Competitive Entrepreneurship Vocabulary: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackBK / 05Business Management & Admin. VocabularyRYAN SHANKAR
Books
Business Management & Admin. Vocab: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackBK / 06Personal Financial Literacy VocabularyRYAN SHANKAR
Books
Personal Financial Lit. Vocabulary: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackBK / 07Business Management & Admin. KPIsRYAN SHANKAR
Books
Business Management & Admin. KPIs: Performance Metrics for Business Growth and Competition
Competitive Business KPIs · Kindle, hardcover & paperbackBK / 08Business Admin. Core VocabularyRYAN SHANKAR
Books
Business Admin. Core Vocabulary: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackBK / 09Competitive Entrepreneurship KPIsRYAN SHANKAR
Books
Competitive Entrepreneurship KPIs: Performance Metrics for Business Growth and Competition
Competitive Business KPIs · Kindle, hardcover & paperbackBK / 10Competitive Marketing KPIsRYAN SHANKAR
Books
Competitive Marketing KPIs: Performance Metrics for Business Growth and Competition
Competitive Business KPIs · Kindle, hardcover & paperbackBK / 11Competitive Marketing VocabularyRYAN SHANKAR
Books
Competitive Marketing Vocabulary: Essential Terms and Strategies for Business and Marketing Success
Competitive Business Vocabulary · Kindle, hardcover & paperbackJR / 01Add a journal articleRYAN SHANKAR
Journal articles
Add a journal article
Journal · Volume · YearCF / 01Add a conference paperRYAN SHANKAR
Conference work
Add a conference paper
Conference · Location · YearWR / 01Prepping for a Record Year: Building Your Chapter’s Calendar with PurposeRYAN SHANKAR
Other writing
Prepping for a Record Year: Building Your Chapter’s Calendar with Purpose
DECA Direct · August 1, 2025WR / 02Behind the Booth: How to Run an Impactful DECA Recruitment TableRYAN SHANKAR
Other writing
Behind the Booth: How to Run an Impactful DECA Recruitment Table
DECA Direct · August 7, 2025WR / 03How to Track Your Chapter’s Summer Goals Like a ProRYAN SHANKAR
Other writing
How to Track Your Chapter’s Summer Goals Like a Pro
DECA Direct · July 15, 2025WR / 04How to Host the Ultimate First DECA Meeting: Engage, Excite, EnrollRYAN SHANKAR
Other writing
How to Host the Ultimate First DECA Meeting: Engage, Excite, Enroll
DECA Direct · August 18, 2025A SMALL INTRODUCTION
What brings you here?
Choose a visitor category to shift only the order in which sections appear. Nothing is hidden, rewritten, tracked, or transmitted.