Paul Osetinsky

About

Welcome! As a software engineer based in NYC with a decade of experience, my specialty lies in backend architecture and cloud infrastructure. I also strive for a wide technical and creative knowledge base, which I believe is essential for prototyping innovative software that merges diverse ideas.

Building creative media across the stack is what I love: servers, databases, DevOps, blockchain, real-time communication, audio, video, frontend, design, and a touch of animation. I find the multi-armed bandit problem an essential guide as I navigate my work, a constant balance of trying new things (breadth) and focusing narrowly on problems that require expertise (depth).

Currently, I'm interested in autonomous systems that utilize generative AI in ways beyond just content creation. I also run a small consultancy called Po Studio focused on helping builders use AI to craft new kinds of products and automate their development. If you're seeking feedback on your project, I'd love to hear from you.

Work

  • 1,989 Sisters: unique NFT collection that bridges physical art with algorithmic collage. I collaborated closely with my friend and artist Blair Breitenstein to write code for generative art, custom logic for smart contracts deployed to the Ethereum blockchain, and systems for storing assets persistently on the decentralized IPFS. $7m in sales volume. Paris even bought one!
  • Upward: iOS app that helped freelancers track their taxes and save for retirement. I joined as founding CTO, building the backend/infrastructure, taking on much of the UI/UX, and overseeing a small team of iOS/QA engineers. We provided an interface that synced financial accounts, analyzed transactions to surface tax-saving deductions, and calculated tax liabilities in real-time (integrations with Plaid, April, and OpenAI). Upward didn't gain enough traction and closed down in early 2024, but I'm still interested in the idea of harnessing LLMs for transaction analysis and narrative insights for personal finance.
  • Dots: built beautiful puzzle games played by millions of people. At Dots, I tackled difficult technical problems and architected several large systems for their flagship game, Two Dots. I also prototyped real-time multiplayer games and explored the use of GANs for generative puzzle creation.
  • Vimeo: I joined Vimeo after they acquired VHX, a startup I had recently joined. I focused on backend engineering to develop tools for independent video creators, building integrations with video transcoding services, migration tools, user authentication and account management systems, and targeted push notifications.

Projects

  • Riddler: a daily word game that crafts its own puzzles, using AI to generate clues in the form of riddles and images. I chose a simple layout similar to Wordle to explore agentic systems for automating game development and content curation.
  • Awestruck: real-time, server-driven music synthesis with SuperCollider, WebRTC, and GStreamer. Applications for human + AI collaborative music composition.
  • Viqi: syncs a video to an input song. Uses audio beat detection and scene extraction to subtly change video playback speeds, ensuring that scene changes fall on musical beats. Relevant for combining generative video and music.

Education

  • Dartmouth College | M.A. in Computer Music (2010)

    In my two years at Dartmouth, I used audio source separation toolkits to develop a novel, subtractive synthesis method. The resulting aural experience is analogous to seeing a blurred image coming into focus, or an object being shattered into pieces. I used this technique to compose original music with SuperCollider, Audacity, Max/MSP, and Ableton Live, performing my work live with ambisonic, multi-channel sound systems referenced in this paper authored by my thesis advisors. I also dabbled in less artsy stuff, and had fun helping undergrads produce EDM in Music 9 (Music & Technology).

  • Vanderbilt University | B.Sc. in Economics (2008)

    In addition to my economics major, I did significant coursework in CS (Java, signal processing), as well as music history, theory, composition, and piano performance of works by Chopin, Prokofiev, Rachmaninoff, Scriabin, Satie, and others. Served as a teaching assistant for Econ 100 (macro) and Econ 101 (micro).

Reach Out

I'm love collaborating on interesting, ambitious projects. If you're seeking feedback on your idea or product, don't hesitate to contact me.