DeepSeek is launching an unprecedented hiring push in Hangzhou, aiming to recruit about 150 senior backend engineers to overhaul and scale its backend systems as user and compute workloads surge.

The platform development track will turn research workflows into platform capabilities to speed up iteration, while the optimization track targets high-performance improvements that could push hardware limits.

The backend tracks span six areas—from large-model research platform to data engineering and the DeepSeek API—designed to support tens of millions of daily active users with low-latency model services.

The Agent Elastic Computing R&D effort centers on DSec, a production-grade sandbox for agents built from three Rust components (APIServer, Edge, Watcher) atop a 3FS file system, aimed at managing hundreds of thousands of concurrent sandbox instances.

In addition to hiring, DeepSeek is expanding its infrastructure to support models, agents, and APIs, signaling a broad backend overhaul across the company.

Specifics of the upcoming IPO remain private, with discussions ongoing about timing, amount to be raised, and target valuation.

The broader AI IPO landscape shows rivals listing in Hong Kong and funding rounds shaping valuation dynamics, underscoring competitive pressures even as DeepSeek pursues its own listing path.

DeepSeek has drawn major backers and signals a push toward substantial funding rounds, with prior rounds featuring large commitments from founders and strategic investors, potentially paving the way for a Star Market or Hong Kong listing.

Fresh capital is being sought to fund computing infrastructure, model development, and talent retention amid competition from both Chinese and U.S. AI firms.

Applicants can apply without selecting a specific track; DeepSeek will assign direction based on background and preferences, reflecting a flexible hiring approach.

The agent lifecycle emphasizes rapid idea-to-validation cycles, keeping researchers close to frontline workflows and enabling efficient development across the six backend tracks.

DeepSeek plans to overhaul the entire system stack to improve reliability and efficiency, signaling ambitious, unprecedented engineering efforts.