AI Native Agent Engineer — Trading

About BlockRun

BlockRun is the routing and payment layer for AI agents. One gateway for LLM inference, real-time market data, and execution APIs — every major model from a dozen providers, prediction markets, crypto + equity + FX feeds, metered per call. No API keys. No subscriptions.

We settle millions of API calls on-chain (see /live) and won the Circle USDC OpenClaw Hackathon — Agentic Commerce track.

The Role

You will design and ship autonomous trading agents on top of the BlockRun stack — agents that read live markets, reason with frontier LLMs, and execute on-chain through a single x402 payment surface. You'll own the agent loop end-to-end: signal ingestion, model routing, risk controls, position management, settlement.

This is a role at the intersection of AI agents and quantitative trading. You bring real trading reps — sell-side desk, prop, market making, DeFi MEV, quant hedge fund, or systematic crypto fund — and pair that with strong agent engineering instincts. The product won't be built by someone who has only read about markets.

This is an early-stage startup. You'll ship code to production daily, talk to users directly, and have outsized impact on the agentic-trading category we're defining.

Under the hood it is USDC pay-per-call over x402: the wallet is the account, so an agent you build here pays for its own inference and data with nothing to sign up for and nothing to top up.

What You'll Do

• Agent loop: Build the core perceive → reason → act → reconcile loop that powers BlockRun trading agents across crypto, prediction markets, and (eventually) equities/FX
• Strategy primitives: Implement reusable agent strategies — momentum, market making, arbitrage across venues, prediction-market resolution arb, news-reactive directional trades — and the backtest harness behind them
• Market data ingestion: Wire BlockRun's data endpoints (Polymarket, on-chain DEX, crypto candles, equities, FX) into a unified market state the agent can reason over
• Execution + risk: Sizing, slippage controls, kill switches, exposure limits, PnL accounting. Production agents have to survive — paper-PnL doesn't ship
• Model routing for trading: Pick the right LLM per decision — fast Haiku/Flash for tick-level reactions, deep Opus/o3 for thesis generation — and tune the routing as new models drop
• On-chain execution: USDC settlement on Base + Solana, smart wallet flows, transaction simulation, MEV-aware order placement
• Observability: Build the dashboards a trader would actually use — live positions, agent reasoning traces, decision quality metrics, drawdown alerts

Tech Stack

You Should Be

• Real trading background — non-negotiable. Sell-side, prop, market making, quant, DeFi MEV, or systematic crypto. You've sized positions with real money on the line and you understand why slippage and fees eat naive strategies alive
• AI native: You use AI coding tools daily (Claude Code, Cursor, Copilot). You ship faster because of them. You've built or seriously prototyped at least one agent before this role
• Strong systems engineer: You can build the trading dashboard in the morning and debug a websocket reconnect storm in the afternoon. Latency budgets, idempotency, and reconciliation aren't new concepts
• Crypto-native: You've held a wallet, signed a transaction, watched a position get rugged. You don't need to be a Solidity expert, but on-chain execution shouldn't intimidate you
• Adversarial-minded: Markets are adversarial; agents are jailbreakable. You think about how the system fails, gets gamed, or loses money — before it does

Nice to Have

• Built a profitable systematic strategy (any asset class)
• Hands-on MEV, on-chain market making, or prediction-market trading experience
• Prior agent / multi-agent system work in production
• Open-source contribution history
• Familiarity with x402 or other agent-payment protocols

Compensation

Competitive. Equity included. All AI tools covered. PnL upside on agent strategies you build is on the table.

Details

Apply

Email hello@blockrun.ai with:

1. A trading project, strategy, or agent you've built — link to repo, writeup, or PnL evidence
2. The market(s) you know best and why
3. Your GitHub handle

No formal resume required. Show us what you've shipped — and what it made or saved.