DeepSeek is forming a new Harness team to build Code Harness from the ground up, and it posted two job listings this week on X. The open roles are Harness Product Manager and Harness R&D Engineer, and candidates are required to be based in Beijing. The job post references Claude Code, and DeepSeek aims to compete with Anthropic and OpenAI.
This week DeepSeek posted two job listings on X, shared by Deli Chen, announcing the formation of a new Harness team to build Code Harness from the ground up. The two open positions are Harness Product Manager and Harness R&D Engineer, presented as the team’s initial hires. Both postings require candidates to be based in Beijing and explicitly list candidate experience with Claude Code, Codex, Cursor, GitHub Copilot, Manus, and OpenClaw, enumerating the specific systems the company seeks familiarity with. The job post also presents the team’s internal formula, “Model + Harness = Agent,” framing how the group describes its technical approach.
DeepSeek’s internal approach to building their Code Harness is summed up by their formula “Model + Harness = Agent.” This methodology frames the creation of intelligent agents through the synergy of their latest model, DeepSeek V4, which functions natively within Claude Code.
In developing their own Code Harness software layer, DeepSeek is positioned to rival established platforms, specifically aiming to compete with Anthropic’s Claude Code and OpenAI’s Codex. By integrating their technology into the Harness, DeepSeek seeks to establish a robust software platform that enhances and maximizes the capabilities of their AI models, further affirming their strategic positioning in the competitive AI industry.
DeepSeek is building Code Harness from the ground up and has formed a Beijing-based Harness team focused on software tooling for agentic coding. The project references Claude Code, involves DeepSeek V4, and identifies Anthropic and OpenAI as comparable platforms.
