DeepSeek is significantly expanding its engineering team to handle growing system complexity. On September 7, the company opened approximately 150 vacancies for senior engineers with 2 to 10 years of experience, as well as recent graduates. The recruitment focuses on server-side development and Agent elastic computing, covering areas such as research platforms, API services, and data engineering.
This marks the second large-scale hiring wave in two months. In June, the company had previously aimed to double the size of all departments, including full-stack development, AI core systems, and deep learning research. Unlike the June drive, the current recruitment specifically targets backend systems and Agent infrastructure, omitting model research roles.
The expansion is driven by a rapid increase in data volume, machine and container counts, and user requests. DeepSeek's Agent elastic computing platform, DSec, already hosts thousands of sandbox environments for next-generation model iteration. The company needs to upgrade and potentially rewrite backend systems to manage resource scheduling, environment isolation, and concurrency stability.
Additionally, DeepSeek has lowered prices for its Flash series API. During off-peak hours, the price for cache-hit input per million tokens has dropped by 60% to ¥0.02 from ¥0.05. Cache-miss input prices fell by approximately 33% to ¥1 from ¥1.5, and output prices decreased by 11% to ¥4 from ¥4.5. Peak pricing remains twice as high as off-peak rates.
The company has also launched testing for the V4.1 Flash intermediate version, which is described as more capable, faster, and cheaper. This recruitment and pricing strategy reflects DeepSeek's transition from a model-focused startup phase to a stage of managing large-scale user requests and complex infrastructure construction.
