attention head AI jobs Alerts Saved
Companies › DeepSeek

DeepSeek

deepseek.com · AI-native company · #40 in the Attention Head AI Index

Open frontier models

Follow DeepSeek

DeepSeek's new roles by email, the morning after we find them. Companies you follow come first in your alert email, with how their hiring moved that week. Narrow it, or keep every role.

How often
We email a link to confirm it.
Open roles
33
AI roles
17
New in 7 days
0
Closed in 7 days
0
Last verified
17 min ago

Open roles

Infrastructure & Platform 6

Data 5

AI/ML Engineering 4

Research 4

Software Engineering 4

Operations & G&A 3

Product 2

People & Recruiting 1

Other roles 4

AI work areas

Trains modelsGPU & kernels

Read from 19 AI roles open now or closed in the last 90 days, 9 naming a work area: 2 name training work (pre-training, post-training, GPU kernels) and 8 name building on models (agents, retrieval).

What DeepSeek's AI roles ask for: how many of its 17 open AI roles name each work area, and (+/−) how many naming it were opened and closed in the last 90 days. Read from the postings' text: what they mention, not everything the team does.

Work area
Agents 7/17 +7 −0GPU & kernels 2/17 +2 −0Retrieval & recommendations 1/17 +1 −0

Index score: 52.8

ComponentValueBasis
Scale (35%)0.45size
AI hiring (25%)0.5117 distinct open AI role titles
Momentum (20%)0.223 role titles posted in 30 days, AI or not
Centrality (20%)1.00frontier

How the score works: methodology.

Models

DeepSeek's most-downloaded public models on Hugging Face.

ModelTaskDownloads, 30 days
DeepSeek-V3.2deepseek-ai/DeepSeek-V3.2 Text generation 10.4M
DeepSeek-OCRdeepseek-ai/DeepSeek-OCR Image text to text 1.8M
DeepSeek-R1deepseek-ai/DeepSeek-R1 Text generation 1.6M
DeepSeek-V4-Flash-0731deepseek-ai/DeepSeek-V4-Flash-0731 Text generation 1.6M
DeepSeek-V4.1-Flashdeepseek-ai/DeepSeek-V4.1-Flash Image text to text 1.4M
DeepSeek-V3deepseek-ai/DeepSeek-V3 Text generation 1.4M
DeepSeek-V4-Flashdeepseek-ai/DeepSeek-V4-Flash Text generation 1.2M
DeepSeek-R1-0528-Qwen3-8Bdeepseek-ai/DeepSeek-R1-0528-Qwen3-8B Text generation 563K
DeepSeek-V4.1-Flash-1017deepseek-ai/DeepSeek-V4.1-Flash-1017 Image text to text 505K
DeepSeek-R1-Distill-Qwen-1.5Bdeepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B Text generation 424K
DeepSeek-R1-Distill-Qwen-32Bdeepseek-ai/DeepSeek-R1-Distill-Qwen-32B Text generation 369K
DeepSeek-V3.1deepseek-ai/DeepSeek-V3.1 Text generation 318K
Is a model wrong or missing?
What's wrong?

Papers

Recent research by DeepSeek's people: papers its Hugging Face organisation lists as its own, and papers whose authors give DeepSeek as their affiliation. Want to work with them? See its 33 open roles.

PaperPublishedCited
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compressionlisted by DeepSeek on Hugging Face Sep 2026 0
A Programming Paradigm for Spatiotemporal ComposabilityarXiv · listed by DeepSeek on Hugging Face Aug 2026 19
DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive GenerationarXiv · listed by DeepSeek on Hugging Face Jul 2026 110
DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligencelisted by DeepSeek on Hugging Face Apr 2026 0
DualPath: Breaking the Storage Bandwidth Bottleneck in Agentic LLM Inferencelisted by DeepSeek on Hugging Face Feb 2026 0
DeepSeek-OCR 2: Visual Causal Flowlisted by DeepSeek on Hugging Face Jan 2026 0
Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Modelslisted by DeepSeek on Hugging Face Jan 2026 0
mHC: Manifold-Constrained Hyper-Connectionslisted by DeepSeek on Hugging Face Dec 2025 0
DeepSeek-V3.2: Pushing the Frontier of Open Large Language Modelslisted by DeepSeek on Hugging Face Dec 2025 0
DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoninglisted by DeepSeek on Hugging Face Nov 2025 0
DeepSeek-OCR: Contexts Optical Compressionlisted by DeepSeek on Hugging Face Oct 2025 0
Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectureslisted by DeepSeek on Hugging Face May 2025 0

All of DeepSeek's papers

Is a paper wrong or missing?
What's wrong?

Where this data comes from

Jobs come from DeepSeek's careers_page job board. Careers page: https://talent.deepseek.com/.

How we found the job board