Senior Software Engineer, Auctions

Alt is unlocking the value of alternative assets, starting with the $5 B trading-card market. We let collectors buy, sell, vault, and finance their cards in one place and we are backed by leaders at Stripe, Coinbase, Seven Seven Six, and pro athletes like Tom Brady and Giannis Antetokounmpo. Our next frontier is real-time pricing at scale—the Alt Value that powers every trade, loan, and product on the platform.

The Role

The last ten seconds of an auction are the whole product. Hundreds of people who have been watching one card for a week all arrive at once, every one of them refreshing, every one of them bidding.

That's this job. You build and scale the high-concurrency bidding and listing systems behind nine figures of annual GMV, and you build the surfaces on top of them. Not one or the other. The low-latency backend logic that resolves a bid and the responsive real-time UI where a collector watches it land are the same feature, and they're both yours.

The number this team is measured on: auction GMV. Every system here either moves it or protects it.

What You'll Own

The auction engine. Build and scale the high-concurrency bidding and listing systems that power nine-figure annual GMV. Thundering herd in the closing seconds is a design constraint, not an incident.

AI-powered liquidity. Develop the next generation of features that use AI to give consignors and sellers instant valuations and upfront liquidity. Being wrong here costs money, which is what makes it interesting.

Personalized discovery. Use data and machine learning to match unique, one-of-one collectibles to the right buyers. Every item is a cold start; that's the problem.

Power-seller tooling. Design and launch the management tools that high-frequency sellers and veteran hobbyists actually need. Watch what they do in spreadsheets today, then delete the spreadsheet.

The whole experience. Backend bid resolution through the responsive UI where the action happens. You own the schema change and the screen it ripples out to.

The bar. Mentor the team and set the standard for code quality, reliability, and performance.

How You'll Use AI Here

We don't want to hear that you use Claude every day. Everyone does. We want to know what you've built with it. Concretely, in this role:

Ship instant valuation into the consignor flow — take an Alt Value, a card's condition and provenance language, and recent comps, and produce an upfront liquidity offer a seller can accept in one tap. This is the flagship AI feature on this team, and it is user-facing

Use vision and language models to enrich listings at scale: parse condition notes, extract attributes from listing photos and descriptions, and turn a messy long tail of one-of-one items into structured data a recommender can actually use

Build a load-and-concurrency harness you can talk to — generate the closing-seconds bid storm, replay it against a branch, and get back where the write path bent, without spending two days wiring a load test

Stand up an agent over auction telemetry that watches close windows and flags anomalies — bid latency creeping, a seller's listings underperforming their comps — before a power seller emails us about it

Put an MCP server over listing, bid, and pricing history so "why did this lot close low" is a sentence, not another one-off query

Use Claude Code or Cursor as the default way you work through a refactor or a migration, not as autocomplete

If you've built agents, wired up connectors, written skills or MCP servers, or shipped internal tooling your team actually depends on — lead with that. Name the artifact, not the tool.

What You Bring

5+ years building and scaling complex applications — Python, Postgres, AWS or comparable cloud infrastructure.

Modern frontend mastery. Deep experience with React, TypeScript, and building high-performance real-time UIs. Real-time is the operative word: this is not a CRUD screen.

Reliability under pressure. A track record of building systems that stay up when the stakes and the traffic are highest. You've been paged during the worst possible ten minutes.