DeepSeek Opens 150 Backend Roles to Rebuild Strained Infrastructure

DeepSeek is recruiting about 150 senior backend and server-side engineers to upgrade, maintain, and rewrite backend systems that a company technical lead said are straining under rapid growth in users, requests, and compute-heavy agent workloads. The openings, posted across DeepSeek’s official recruitment portal in Beijing and Hangzhou, were disclosed on September 7, 2026.

Tianyi Cui, whose verified X profile identifies him as a member of technical staff on DeepSeek’s Harness team, announced the hiring push in a post on X. He wrote that DeepSeek was opening roughly 150 headcount positions for experienced backend and server engineers, citing many new directions, systems, and requirements. In a follow-up post, Cui described the scale of the recruiting round as unprecedented for the company and explained the pressure behind it.

According to Cui, the push stems from a sharp rise in complexity that comes with scale. He wrote that data volumes, machine and container counts, training tasks, evaluation tasks, agent environments, user numbers, and request volumes are all increasing rapidly. That growth, he said, means DeepSeek’s existing backend systems can no longer perfectly meet future demand, requiring the company to bring in large numbers of engineers to upgrade, maintain, and rewrite those systems. Cui framed his explanation as his personal understanding of the reason for the recruiting round.

Roles and Locations

DeepSeek’s official careers portal lists the two positions covered by the recruiting round, along with dozens of other openings. Both of the roles tied to the 150-person expansion are posted in Beijing and Hangzhou under the company’s social-recruitment channel.

The first role, server development engineer, states explicitly that the recruiting round covers approximately 150 people and that work will be based primarily in Beijing, with some positions available in Hangzhou. The listing says DeepSeek welcomes recent graduates and engineers with up to two years of experience, but identifies senior engineers with two to ten years of work experience as the focus of this round, adding that the experience requirement can be relaxed for exceptional candidates.

The server development role is divided into six tracks. These cover a large-model research platform, agent framework components, research-and-development efficiency infrastructure, the DeepSeek API, online services, and data engineering. The listing describes the online-services track as responsible for the architecture and continuous iteration of large-model applications serving tens of millions of daily active users, and the API track as focused on delivering reliable, low-latency model services to global developers.

The second role, agent elastic computing research and development engineer, covers DSec, DeepSeek’s elastic computing platform built for agents. The listing describes DSec as hosting thousands of sandbox environments and supporting the iteration of DeepSeek’s next-generation models. It is split into a platform-development-and-maintenance track and a lower-level systems-tuning track, and lists work spanning operating systems, virtualization, storage, networking, and control-plane services.

The DSec Platform

The DSec job description says the platform must modify its entire system stack to improve reliability and efficiency, from the operating system and virtual machines through network and storage layers up to application-level scheduling and control-plane services. It adds that many of the required tasks have not been attempted before and have no ready-made solutions to copy.

The architecture described in the listing matches the one DeepSeek documented in its DeepSeek-V4 technical report. In that report, DeepSeek describes DSec, short for DeepSeek Elastic Compute, as a production-grade sandbox platform for agentic AI built from three Rust components: an API gateway called Apiserver, a per-host agent called Edge, and a cluster monitor called Watcher. The components connect through a custom RPC protocol and run on top of DeepSeek’s 3FS distributed file system. According to the report, a single DSec cluster manages hundreds of thousands of concurrent sandbox instances in production.

Cui’s posts did not disclose compensation, and neither job listing states a salary range. The listings note that candidates can apply without selecting a specific track, with DeepSeek assigning direction based on background and preference.