Tech-tag co-occurrence Every AI role gets tagged from a 35-term controlled vocabulary spanning agent / eval / training / inference / modality clusters. Tags that frequently appear together in the same JD pull each other close; thicker edges = more co-occurrences. The clusters that emerge organically are the real sub-disciplines of applied AI engineering right now.
Showing 115 tagged AI roles in Defense. Layout is a Fruchterman-Reingold force simulation, run server-side to convergence.
agent_orchestration ↔ agent_research (3 JDs) agent_orchestration ↔ rl_robotics (4 JDs) agent_orchestration ↔ embodied_ai (17 JDs) agent_orchestration ↔ synthetic_data (2 JDs) agent_research ↔ rl_robotics (1 JDs) agent_research ↔ embodied_ai (2 JDs) agent_research ↔ synthetic_data (1 JDs) embodied_ai ↔ rl_robotics (4 JDs) rl_robotics ↔ synthetic_data (1 JDs) embodied_ai ↔ synthetic_data (1 JDs) agent_orchestration ↔ multimodal (5 JDs) agent_orchestration ↔ evals (6 JDs) evals ↔ multimodal (4 JDs) inference_infra ↔ vision (3 JDs) model_serving ↔ vision (3 JDs) inference_infra ↔ model_serving (16 JDs) agent_orchestration ↔ tool_use (5 JDs) fine_tuning ↔ model_serving (1 JDs) agent_orchestration ↔ fine_tuning (1 JDs) agent_orchestration ↔ model_serving (10 JDs) agent_orchestration ↔ multi_agent (12 JDs) multi_agent ↔ rl_robotics (1 JDs) embodied_ai ↔ multi_agent (2 JDs) agent_research ↔ evals (1 JDs) embodied_ai ↔ evals (1 JDs) agent_orchestration ↔ llm_observability (4 JDs) evals ↔ tool_use (2 JDs) llm_observability ↔ tool_use (3 JDs) multimodal ↔ tool_use (2 JDs) evals ↔ llm_observability (1 JDs) llm_observability ↔ multimodal (1 JDs) evals ↔ model_serving (2 JDs) evals ↔ inference_infra (2 JDs) agent_orchestration ↔ rag (2 JDs) agent_orchestration ↔ vector_db (2 JDs) multi_agent ↔ rag (1 JDs) multi_agent ↔ vector_db (1 JDs) rag ↔ vector_db (3 JDs) agent_orchestration ↔ inference_infra (3 JDs) inference_infra ↔ multi_agent (2 JDs) model_serving ↔ multi_agent (2 JDs) rag ↔ tool_use (1 JDs) tool_use ↔ vector_db (1 JDs) llm_observability ↔ rag (2 JDs) llm_observability ↔ vector_db (2 JDs) agent_research ↔ model_serving (1 JDs) llm_observability ↔ synthetic_data (1 JDs) model_serving ↔ rag (1 JDs) model_serving ↔ vector_db (1 JDs) llm_observability ↔ model_serving (1 JDs) inference_infra ↔ rag (1 JDs) inference_infra ↔ vector_db (1 JDs) inference_infra ↔ llm_observability (1 JDs) rag ↔ vision (1 JDs) vector_db ↔ vision (1 JDs) llm_observability ↔ vision (1 JDs) agent_orchestration ↔ vision (1 JDs) agent_orchestration ↔ frontier_research (1 JDs) multimodal ↔ vision (1 JDs) frontier_research ↔ vision (1 JDs) frontier_research ↔ multimodal (1 JDs) multimodal ↔ training_infra (1 JDs) multimodal ↔ synthetic_data (1 JDs) inference_infra ↔ multimodal (1 JDs) model_serving ↔ multimodal (1 JDs) evals ↔ training_infra (1 JDs) synthetic_data ↔ training_infra (1 JDs) inference_infra ↔ training_infra (1 JDs) model_serving ↔ training_infra (1 JDs) evals ↔ synthetic_data (1 JDs) inference_infra ↔ synthetic_data (1 JDs) model_serving ↔ synthetic_data (1 JDs) Agent orchestration
N=63 JDs
Top co-occur: Embodied AI ×17 · Multi-agent ×12 · Model serving ×10 Agent orchestration Agent research
N=4 JDs
Top co-occur: Agent orchestration ×3 · Embodied AI ×2 · RL robotics ×1 RL robotics
N=4 JDs
Top co-occur: Agent orchestration ×4 · Embodied AI ×4 · Agent research ×1 Embodied AI
N=24 JDs
Top co-occur: Agent orchestration ×17 · RL robotics ×4 · Agent research ×2 Embodied AI Synthetic data
N=3 JDs
Top co-occur: Agent orchestration ×2 · Agent research ×1 · RL robotics ×1 Multimodal
N=6 JDs
Top co-occur: Agent orchestration ×5 · Evals ×4 · Tool use ×2 Evals
N=9 JDs
Top co-occur: Agent orchestration ×6 · Multimodal ×4 · Tool use ×2 Vision
N=12 JDs
Top co-occur: Inference infra ×3 · Model serving ×3 · RAG ×1 Inference infra
N=17 JDs
Top co-occur: Model serving ×16 · Vision ×3 · Agent orchestration ×3 Inference infra Model serving
N=24 JDs
Top co-occur: Inference infra ×16 · Agent orchestration ×10 · Vision ×3 Model serving Tool use
N=5 JDs
Top co-occur: Agent orchestration ×5 · LLM observability ×3 · Evals ×2 Fine-tuning
N=1 JDs
Top co-occur: Model serving ×1 · Agent orchestration ×1 Multi-agent
N=35 JDs
Top co-occur: Agent orchestration ×12 · Embodied AI ×2 · Inference infra ×2 Multi-agent LLM observability
N=5 JDs
Top co-occur: Agent orchestration ×4 · Tool use ×3 · RAG ×2 RAG
N=3 JDs
Top co-occur: Vector DB ×3 · Agent orchestration ×2 · LLM observability ×2 Vector DB
N=3 JDs
Top co-occur: RAG ×3 · Agent orchestration ×2 · LLM observability ×2 Frontier research
N=1 JDs
Top co-occur: Agent orchestration ×1 · Vision ×1 · Multimodal ×1 Training infra
N=1 JDs
Top co-occur: Multimodal ×1 · Evals ×1 · Synthetic data ×1 18 tags · 72 co-occurrence edges · min edge weight 1. Bubble area ∝ JDs containing tag · edge thickness ∝ co-occurrence count. Hover any node for top-3 partners; click to see the JDs. Tag velocity · last 4 weeks vs prior 4 Which technologies are hot, which are cooling. Sparkline = 12 weeks of unique roles tagged with each term, last bar on the right is this week. Sorted by absolute pickup. Tags with under 10 lifetime mentions are hidden as noise.