MIDDLE AI SOFTWARE ENGINEER

We are looking for a Middle AI Software Engineer (Python / C++) who will help us scale from "an engineer improves business results while spending thousands of dollars daily on tokens" to "multiplies the business impact of the entire team—even when spending $10k+ daily on tokens."

About the role

You give an agent a task, and within twenty minutes, it delivers "working" code that compiles and passes tests. However, hidden within are subtle bugs, invented APIs, and suboptimal solutions. Quickly spotting these, understanding why the agent made those choices, and refining the code to the desired state—this is the skill we seek. Agents accelerate work significantly, but the responsibility for every line of code remains with the engineer.

We build trading infrastructure for high-frequency trading (HFT): a large system that must operate stably with minimal latency during critical moments under loads 10x higher than usual. To achieve this, we use our own user-space network stack, sub-microsecond IPC, optimized data formats, and more.

A Middle AI Software Engineer here is a strong Python and C++ engineer who leverages agent tools to do more individually and boost the productivity of the entire team. This is not a "prompt engineer" role—it’s a software engineering position where AI skills complement a solid engineering foundation rather than replace it.

Responsibilities

• Build agent workflows: assemble pipelines where agents decompose tasks, break them down, and verify each other’s results. Select what and how to delegate, refining these processes until the whole team uses them.
• Integrate LLMs into services where justified, using tool use and structured output via API with execution control. Understand which tasks should (or shouldn’t) be delegated to models.
• Tackle full-scale HFT tasks with agents as a multiplier: connecting new exchanges, researching exchange infrastructure, and optimizing latency.
• Develop internal agent tooling: agent rules and instructions tailored to our codebase, custom tools for agents, MCP servers, and evals—infrastructure that makes agent use efficient across the team.

Requirements

• Daily use of agent coding tools (Claude Code, Codex, Cursor) on real tasks, with a track record of delivering commercial results.
• Ability to identify where agents accelerate work and where they introduce hidden technical debt, understanding why outputs are efficient or inefficient—at the level of caches, pipelines, and memory models.
• Quickly spot subtle bugs, hallucinations, and issues in generated code, taking full responsibility for the final output alongside hand-written code.
• Engineering-level understanding of LLMs: context, tokens, causes of hallucinations, tool use, and how to measure and control output quality.
• Strong Python skills and the ability to confidently work in a C++ codebase.
• Understanding of OS and hardware (CPU, RAM, NIC), network protocols (Ethernet, UDP, TCP), and the basics of multithreaded programming.

While real skills matter more than formal experience, engineers with 3+ years of commercial server development (including production C++ experience) typically excel in these tasks.

Nice to have

• Experience building agent pipelines or multi-agent orchestration, MCP servers, or evals.
• Integration of LLMs into services via API: tool use/function calling, structured output, streaming, and cost management.
• Customization of AI tools for large codebases: rules, custom tools/hooks, and indexing.
• HFT experience: order routing, market data processing, or exchange research.

Benefits

• Unlimited budget for tokens: spend $10k+ daily if needed to achieve quality results.
• Personal responsibility for outcomes: your decisions go into production and impact PnL, with growing responsibility and freedom in approach.
• Profit sharing: your income reflects your personal contribution to the team’s results (avg. +50–100% of annual income).
• A strong team: engineers from tier-1 companies, industry veterans, and winners/medalists of math, programming, and physics olympiads (IMO, IOI, ВсОШ, ICPC semifinalists).
• Comprehensive employee care:
  • Health monitoring and support for key health metrics.
  • Dental-inclusive health insurance (DMS).
  • Corporate events worldwide.
  • Monthly sports and wellness activities.
  • Office meals from top restaurants.
• Deep work environment: no meetings for the sake of meetings, short feedback loops, and performance coaching to boost personal productivity.
• Tasks that directly impact PnL: no "unimportant" work—we structure tasks to maximize your efficiency.