# Thomas Yee (tyeetale)

> Operator and builder. AI systems, data infrastructure, and products.
> Currently helping rockets pencil out at Blue Origin.
> The rest of my time goes to shipping AI that gives people superpowers.

## Identity

- Name: Thomas Yee
- Handle: tyeetale
- Location: United States
- Archetype: Three-time founder, AI product builder, and data engineer behind a multi-billion-dollar rocket program
- Tagline: Consistent output + improvement
- Summary: Built recommendation systems before transformers were mainstream, launched an AI travel planner before agentic workflows became popular, and now builds agents that reason and act on behalf of users.

## Current Role

- Blue Origin — Financial Data Engineer, FP&A New Glenn (August 2024–present)
- Constructs the ~$5.4B New Glenn Annual Operating Plan across 12 rocket subsystems
- Reduced 30+ hours/month of reporting effort by architecting a Databricks-based data system with 10+ ETL workflows
- Increased financial ownership across 320+ budget owners with self-serve metrics and automated monthly reporting
- Surfaced $200M+ in stale purchase orders across 1,000+ suppliers via Non-Labor KPI suite
- Won Blue Origin's corporate AI hackathon (20+ teams) by building an AI agent using MCP servers and RAG pipelines over enterprise knowledge (Databricks, GitLab, SharePoint, Jira)

## Side Projects (Active)

- Mosslayer (https://mosslayer.com) — Co-Founder (Apr 2025–present). One of the first consumer-facing agentic payment platforms. AI evaluates financial context, recommends actions, and executes purchases autonomously over crypto rails. Reached $5K GMV in first month. Owns product strategy and agent architecture from zero to production.
- Onsite (https://getonsiteai.com) — Founder (Apr 2025–present). AI-powered company brains for home renovation contractors. Reduces time managing bids, invoices, projects, and customer relationships. Onboarded 5 contractors in beta. Built from 10+ contractor interviews.

## Past Work

- Tildenn (Jan 2022–Aug 2024) — Founder. One of the earliest AI travel planners. Collapsed trip planning from 2 days to under 30 minutes for 100+ customers. Shipped 10+ major product iterations in 6 months from 50+ user feedback sessions. Secured $15K+ in non-dilutive funding through NYU's Entrepreneurial Accelerator Program. Hybrid system: GPT-3 + hierarchical clustering + haversine distance + rule engines. Next.js, MongoDB, Lambda.
- Coopsight (Oct 2019–Dec 2021) — Co-Founder/COO. ML matchmaking engine (word2vec, Jaccard similarity, TesseractOCR) generating 100+ M&A and partnership opportunities per fund across 10 VC portfolios. Onboarded 250+ startups. Closed partnership with Sberbank, Russia's largest bank. Directed cross-functional team of 9. React, Node, Firebase, Kubernetes, GCP.
- Metaphor3D (2021–2022) — Text-to-3D asset generation. Three.js web viewer. Pre-Stable Diffusion era exploration of generative AI beyond text.
- Fibes (2020–2021) — UGC influencer marketplace. Loyalty-driven model where creators earn products instead of cash. Incentive design exploration.

## Education

- NYU Shanghai (https://shanghai.nyu.edu) — BS Finance, Data Analysis (2018–2022)
- Sino-US research university designed to produce global citizens
- Dual concentration: Business Analytics & Marketing
- Merit Scholarship
- 22+ countries by graduation

## Career Arc

Finance/data analysis (NYU Shanghai) → ML-driven products (Coopsight) → Generative AI exploration (Metaphor3D) → Solo AI product builder (Tildenn) → Data infrastructure at scale (Blue Origin) → AI products combining all learnings (Onsite, Mosslayer)

## Principles

- Ship, then refine. Working things teach more than perfect plans.
- Systems over features. Build the system that makes features trivial.
- AI is an interface problem. Models commoditize, UX differentiates.
- Data compounds. Every pipeline, connection, and pattern captured compounds over time.
- Cross-discipline advantage. Finance, design, and engineering intersect to produce better products.
- Intent over action. Map what someone needs before what they asked for.
- Consistent output. Show up every day and build. The compounding is in the consistency.
- Use the right tool at the right time. Transformers for language. Classical ML for clustering. Rule engines for constraints.

## Currently Exploring

- Agentic systems: how to build AI agents that reliably act on behalf of users
- Intent mapping: going from "what the user said" to "what they need" to "what action to take"
- Data infrastructure at human scale: surfacing insights at the moment they matter
- AI-native UX patterns: interaction beyond forms and buttons
- Cost of intelligence: architecting systems where the right model runs at the right time for the right cost

## Technical Skills

- AI: LLM agents, agent evaluation frameworks, MCP servers, Amazon Bedrock, RAG, prompt engineering, OCR pipelines, vector search, word2vec, retrieval systems, context engineering
- Engineering: Python, SQL, TypeScript, React, Next.js, Node.js, Databricks, ETL pipelines, AWS, GCP, MongoDB, Kubernetes, Cloudflare Workers, Supabase, Docker
- Product: User research, roadmapping, monetization experiments, Agile, Jira, Figma, forecasting, unit economics
- Tools: Cursor, Warp, Git, Linear, Arc, Vite, TailwindCSS

## Links

- Site: https://tyeetale.vercel.app
- GitHub: https://github.com/tyeetale
- X: https://x.com/tyeetale
- LinkedIn: https://linkedin.com/in/tyeetyee
- Email: tyeetale@gmail.com

## Site Pages

- / — Homepage with intro, timeline, projects
- /about — Principles and current interests
- /projects — Project listing
- /projects/tildenn — AI travel planner case study
- /projects/coopsight — Startup ecosystem matchmaker case study
- /projects/metaphor3d — AI 3D asset generation
- /projects/fibes — UGC influencer marketplace
- /tools — Interactive widgets and experiments
- /graph — Force-directed knowledge graph connecting projects and topics
- /qr — QR code for sharing
- /llms.txt — This file (machine-readable profile)

## Project Connections (Knowledge Graph)

- Tildenn connects to: Coopsight (same founder, travel → data)
- Coopsight connects to: Fibes (same team, pivoted from partnerships to content)
- Metaphor3D connects to: Tildenn (generative AI exploration → AI product)
- Fibes connects to: Coopsight (same team), Mosslayer (incentive design → agent payments)
- Onsite connects to: Blue Origin (AI at scale), Tildenn (AI product building)
- Mosslayer connects to: Fibes (intent mapping), Onsite (AI products)
- Blue Origin connects to: NYU Shanghai (finance → data infrastructure)
- NYU Shanghai connects to: Coopsight (founded during studies)

## Topics

AI, Data, ML, Product Design, Travel, Consumer, Startups, Generative AI, 3D, Marketing, Finance, Payments

## Meta

- This file follows the llms.txt convention for AI discoverability
- Last updated: 2025-06-12
- Site built with: Vite, React, TypeScript, TailwindCSS, MDX
- AI chat available at /graph (powered by Cloudflare Workers AI)
- Graph visualization shows project/topic connections interactively