I build real products, not just decks.

Problem

A new line of pulsed RF plasma generators (AVA 1213, 3013, 6013) needed a product identity, launch kit, and digital presence for SEMICON West 2025.

Result

Built the AVA brand identity, banners, brochures, booth concepts, and LinkedIn plan, supporting 170+ booth demos, 60+ product-info requests, and 20+ partnership leads; then planned the AVA website and content engine for 2026.

Problem

Service businesses lose jobs to missed after-hours calls, overloaded office staff, and voicemails customers won’t leave, while leads slip between phone, SMS, and CRM.

My role

Co-founder, Product. I own product strategy and roadmap, voice agent and call flow design, CRM integrations, positioning and pricing, the website and demo, and customer discovery and sales.

Result

Built a 24/7 AI receptionist that answers, triages, and books calls and hands off to Housecall Pro, Jobber, and ServiceTitan, with a self-serve trial and three usage-based plans from $99 to $225 a month. Now in early access.

Problem

Blue Ocean runs jobs, scheduling, and invoicing in HouseCall Pro, but quoting and personal-spend reimbursements needed custom workflows HCP doesn’t capture automatically.

Result

As Business Operations Intern, gathered requirements from the owner and staff, then shipped a Next.js agentic estimate tool on Fly.io with HouseCall Pro integration and HCP-aligned expense workflows. A staffing model I built identified ~$3,400/month in potential slow-season savings.

Problem

With international marketing rights in Mexico and the UK, the 49ers have room to grow a global fanbase that responds to different content than US fans.

Result

Pitched, as a prospective intern, a creative product and brand strategy: 7 target markets and 12+ content concepts adapted from global soccer, cricket, and basketball brands.

Problem

Deel has Arsenal on the world stage, but global sponsorship drives awareness, not conversations with local employers who need modern HR, payroll, and global hiring.

Result

Proposed, independently, a 6-week Bay Area pilot with one community football club, AI-assisted and human-reviewed targeting, and a single north-star KPI: cost per sales-accepted qualified meeting.

Problem

Can a creator predict engagement before uploading, using only metadata they control such as title, length, and timing?

Result

Modeled 250 MrBeast uploads from the YouTube Data API with tree models and SHAP; duration, Shorts format, and an 18–22h upload window were the top drivers.

Problem

Optimize pricing, fleet deployment, routing, and demand coverage across a constrained distribution network.

Result

Built a Python integer-programming model with sensitivity analysis on price, ingredient cost, and fleet size, producing scenario-based recommendations for operational decisions.

Problem

An AI-native marketing automation platform needed clearer positioning and GTM narratives for buyers and investors.

Result

Defined positioning via market and competitive research; built launch content and investor materials for UC Berkeley SkyDeck Investor Month.

Includes

Statistical Modeling, Data Mining, the Taiwan credit card crisis case study, and additional analytics builds on GitHub, plus the full interactive decks behind these case studies.