Music taught me discipline, design taught me structure, and writing taught me how to tell a story. I bring all three to engineering.
About
Peter Markoski.
I got my start in technology and music in Washington, D.C., making spaces for people to hear each other’s music. These days I am based in Los Angeles, blending my love for both mediums.
Selected work / 01
Coinbase
Redesigning how people place a trade.
Role
Lead frontend engineer
Status
Current role
Order entry, shared foundations, and agentic AI
I owned the rewrite of Coinbase’s consumer derivatives order forms, working with design to create a cleaner, single-page trading experience.
Underneath the interface, I addressed fragmented state management, feature-specific branches, and deeply passed props. The rewrite established shared components and core logic used across web and mobile, giving other engineers a consistent foundation for new work.
I also protected delivery as scope and competing priorities evolved—negotiating compromises across teams, guiding technical decisions, and maintaining coding standards. The components and logic continue to support derivatives development today.
My work also extends into agentic AI and new product development at Coinbase. I’m building systems of collaborating AI agents and helping establish the technical foundations for emerging products. These initiatives are still in development, with further details remaining internal.
Coinbase’s derivatives order forms had grown through successive feature additions. Feature-specific branches, large GraphQL queries, fragmented state management, and prop drilling made the implementation increasingly difficult to extend. The customer experience also felt dated and cumbersome.
Owning the rewrite
I owned and implemented the order forms within a derivatives frontend team of roughly five engineers. Working with the design team, I helped bring order entry into a cleaner, single-page experience while rebuilding the technical foundation underneath it.
A shared engineering foundation
The rewrite addressed how data and state moved through the forms. I established more deliberate state-management patterns using external React stores and Context providers, with a focus on clearer ownership and reducing the complexity passed through component trees. Shared packages made core logic and reusable components available across web and mobile.
Keeping delivery focused
This work also required keeping delivery focused as priorities changed. Other initiatives competed for engineering capacity, and teams brought strong opinions about updates that intersected with the rewrite. I pushed back when expanding scope threatened delivery, guided technical decisions toward consensus, and negotiated compromises with stakeholders and partner teams.
A foundation that continues to support new work
My responsibility extended beyond completing the initial implementation. I upheld coding standards and established patterns that other individual contributors could build on as derivatives products evolved.
The result was a redesigned order-entry experience supported by a more coherent engineering foundation. The components and shared logic continue to be used across derivatives teams, extending the value of the rewrite into subsequent development.
Agentic AI and new product development
Alongside derivatives trading, I’m building agentic AI systems at Coinbase, including swarms of collaborating agents, and contributing to the engineering foundations of new products.
My work spans technical planning, hands-on development, and coordination across teams as these initiatives take shape. These projects remain in development, so I’m keeping their product details and internal architecture private.
I’m building Track Tree so musicians in different studios can hear the same passage at the same time. A producer shares a mix by link and hosts synchronized playback. Collaborators can also listen on their own, isolate a stem and leave a timestamped note on the exact sound they want changed.
The next bounce goes into the same project as a new version. Everyone can compare takes and revisit earlier feedback without hunting through file transfers or guessing which export a comment refers to.
Say a vocalist flags a doubled breath just before the chorus. The engineer can open that note, solo the vocal stem and hear the breath in the mix where it was flagged. A later edit might move the chorus by two bars, so the original note stays with its version. I’m designing the first DAW integration around Ableton Live, so an engineer could follow that same note to the passage in the session where they can edit it.
I got my start in electronic music with room.haus, a Washington, D.C., collective. We used WebGL to turn a DJ mix into a space you could explore in your browser.
What changes when the market gets new information?
Role
Research & engineering
Status
Ongoing research
An automated loop for testing trading ideas
I built a research pipeline that takes a trading hypothesis through backtesting, comparison and reporting in one automated run. Python runs each candidate and its baseline against the same market history. Both call the same Rust engine, so I can trace a difference in results back to the strategy change that produced it.
For market sentiment research, I integrated Jev, a fast AI model that screens headlines and summaries as they arrive over RSS. It flags developments such as earnings surprises and guidance changes, giving me a timestamped record of what the strategy could have known as trading unfolded.
How far back should a strategy look after news breaks?
A surprise earnings announcement raises a practical question. Do the last few minutes of trading tell us more than the hour before the news? The RSS collector records when it first sees the headline. Jev classifies the event and its relevance to equities. I’m exploring whether those signals can help select a shorter or longer lookback window for live trading. A fixed window gives me a baseline for testing whether adapting to the news adds value.