Most people know me for one thing. This page is the whole picture: peer-reviewed research, AI products, quant trading systems, club leadership, and the students I teach along the way. Deep dives live on my personal site.
PythonJavaJavaScriptReactNext.jsSQLIBM WatsonNLPLLMs & RAGQuant TradingFull-StackCS Education Research
01 · Research & AI
Published work
I do CS education research at the SJSU CSEd Research Lab under Dr. Ethel Tshukudu — mixed-methods studies on how students actually learn to program. On the applied side, I build AI systems that have to survive real data, not just demos.
SIGCSE TS 2026 · Poster Abstract
Exploring Bilingual Coding for Inclusive Computer Science Learning
Co-authored a mixed-methods study with 60 participants exploring how bilingual programming environments lower the barrier for non-native English speakers learning CS. Contributed across the pipeline: study design, data collection, and analysis. Published in the ACM Digital Library.
An IBM Watson–powered sentiment engine that scores equities from market news. Built end-to-end — ingestion, NLP pipeline, scoring model, and the dashboard on top. A learning project from IBM SkillsBuild; it was never run in production, so there are no audited performance figures to quote.
A free bilingual intro-CS program for high schoolers still learning English, built out of my own SIGCSE research on bilingual coding. Concepts taught in the student's home language first, English keywords attached second. Curriculum, volunteer handbook and a four-language glossary are written and public. No cohort has run yet, no school partnership is signed, and it is not an incorporated nonprofit.
Built AI intake, billing, and accounts-receivable automation for a real freight brokerage: a passing regression suite, the AR and billing pain points automated, 100% local models, $0 API spend. The deepest production AI work I've done — full case study on the site.
An ffmpeg pipeline that turns raw event footage into a finished 9:16 highlight reel — scene detection and audio-energy scoring pick the moments, no editing skill required — plus a gold-and-black local web studio to drive it. Working prototype, not launched.
A community trash map with a quantified "world fixed" impact meter — report a pin, clean it, log an eco action, watch kg of waste and CO₂e climb. Privacy is designed in at the schema: no user GPS stored and coordinates rounded. Working local MVP.
$1.99 flat-fee food delivery built for the SJSU campus: nearby restaurants on one dark-mode map of cravings, delivered by fellow students. The courier jobs board even supports food-barter payouts — accept three slices and a Gatorade as your fee.
A setup screener and no-lookahead backtest engine — then the adversarial audit I ran against my own headline result across 129 symbols and 10 years. It came back marginal: against risk-matched random entries the signal's own contribution was +0.030R, 95% CI [−0.014, +0.077]. Building the test that kills your own idea is the actual skill.
An execution engine where the strategy is forty lines and everything else is crash-safety: deterministic order IDs, a journal claimed before every broker call, linked exits, and idempotent reconciliation. 160 passing tests. Paper mode only — no live capital.
Built and maintain sventclub.org, the digital front door for the Silicon Valley Entrepreneurship Club — events, sponsors, and member funnel. Part of the same push to make the club's events and membership easier to run.
Code is half of it. The other half is rooms full of people: running club operations, managing a chapter's money, and teaching CS to students who remind me why I started.
Off the laptop — San Jose, 2026
SVEC
Technical Ops & Web Lead
Run the technical side of the Silicon Valley Entrepreneurship Club — site, tooling, and event ops. Ran the club's site, tooling and event operations for the year.
Kappa Sigma · SJSU
Board Treasurer
Own the finances for a 50+ member chapter: budgeting, dues, and the books behind 10+ events a year. Real money, real accountability, real spreadsheets.
SJSU CS Dept
CS Tutor & Curriculum Lead
Tutor with Coding Warriors and lead curriculum for CS outreach programs at Bay Area high schools — the same programs my research studies, so what I learn in the lab ships to actual classrooms.
Mathos AI
Campus Ambassador
Represent an AI math-learning platform on campus since early 2025 — running demos, gathering student feedback, and feeding it back to the product team.
04 · Products
Founding-client pilots open · pre-revenue
FreightDesk AI — an AI back office for freight brokers
The AR and billing automation I built inside a real freight brokerage, productized: invoice chasing, POD prep, and inbox AI that runs 100% locally — no data leaves your machine. Founding-client pilots are $999 one-time setup + $249/mo — a part-time AR clerk runs $1,800–2,200 a month. Founding-client launch pricing while I run the first cohort.
Peer-reviewed computing-education research at SIGCSE 2026 · an NLP sentiment-scoring prototype built on IBM Watson · a green regression suite in FreightDesk — all of it built, tested and public. The newest door: FreightDesk founding-client pilots, open now, first slot still unclaimed.
Trading figures on this page are backtest and walk-forward simulation results, not live trading performance. That early headline did not survive my own adversarial re-test and I say so on the page. Hypothetical results do not account for real slippage, fees or drawdown, and past performance does not guarantee future results. No live profit-and-loss figures are claimed anywhere on this site. FreightDesk pricing and cost comparisons are list prices and illustrative math, not a projection of your savings. Nothing here is financial, investment, tax or legal advice.