M87
Shift: RFQ Market Maker
Build & Ship / Fractional CTO

Shift: RFQ Market Maker

We built Shift's RFQ market maker for tokenized equities on Jupiter. Live trading on Solana, with inventory controls and AI-assisted maintenance.

ClientShift
Niche

Crypto / Trading

Timeline

Ongoing

Key ResultLive RFQ trading on Jupiter

The Result

Shift's tokenized equities are trading through a production market maker we built for Jupiter on Solana. It publishes buy and sell prices, manages available inventory, validates swap requests, and signs trades that meet its rules.

We also made the system maintainable through an AI agent. Investigating a rejected swap, implementing a change, running tests, and preparing a release can happen through a conversation, with engineering review and approval before deployment.

The Challenge

Shift needed its own liquidity infrastructure for tokenized equities. A displayed market price was only the starting point. Users needed executable quotes for the amounts they wanted to trade, and the business needed control over the inventory behind those quotes.

The system also had to bridge two different markets: equities with exchange opening hours, and tokens that can trade outside those hours. Pricing, available funds, transaction outcomes, and operator visibility all had to work together.

What We Built

Pricing and inventory

We built a request-for-quote (RFQ) market maker that connects underlying market data to prices published on Jupiter. Quotes account for trade size, available inventory, and whether the underlying market is open. Token amount conversions are handled consistently between pricing, settlement, and the admin interface.

Trade validation and settlement

The maker supports direct swaps and supported multi-leg routes. It checks proposed trades against its pricing and inventory rules before signing, then tracks the on-chain outcome. Rejected requests and failed transactions remain available for investigation rather than disappearing into temporary logs.

Tools for the operating team

The admin interface brings together token controls, pricing, inventory, and trading records. Telegram notifications cover operational events, fills, inventory movements, and rejected quotes. Date-range reports with PDF export give the team a way to review maker activity outside the live dashboard.

Maintenance Through Prompts

The AI agent works on the engineering and operations workflow. It does not decide trade prices or replace the maker's transaction checks.

An operator can ask why a swap was refused or request a change to an alert. The agent can inspect the relevant code and records, explain the cause, implement a fix, and run the checks needed for review. Approved changes can then be deployed and verified through the same conversation.

That workflow depends on documented operating procedures and tests covering pricing, inventory, transaction validation, and settlement. We use those checks to evaluate changes before they reach production. A prompt is a way to request work, not a substitute for verification.

The Engagement

The engagement combines Build & Ship with Fractional CTO. We provide technical direction, make architecture and release decisions with Shift, and build and maintain the system ourselves.

That includes taking the maker from implementation through launch, then continuing to develop route support, reporting, and operational alerts. Real trading activity and user feedback guide what we work on next.

Shift has a live RFQ market maker and an operating workflow its team can use. We continue to refine the system against actual trades and user feedback.

Technologies Used

TypeScriptNode.jsSolanaJupiterPostgreSQL

Tell us what you need.

Whether it's a product, a team transformation, or a developer - we'll get back to you with an honest answer on whether we're the right fit.