T
TECHNOLOGY ARCHITECT/ Senior AI Engineer (Applied AI)
TechDigital Group
5 hours ago
Full-time
On-site
Blaine, Minnesota, United States
What are the top 3 skills required for this role?
Agentic Workflows & Memory Systems
Stateful Orchestration: Building and debugging production-grade, cyclic multi-agent workflows and state machines.
Context & Memory Engineering: Implementing multi-layered memory architectures including short-term memory for active turn execution, long-term memory for cross-session state persistence, and episodic summarization to manage token context windows.
Tool Call Management: Designing dependable function-calling patterns equipped with automated retry logic and self-correction handlers.
Data & Retrieval Infrastructure
Vector & Relational Storage: Managing relational metadata schemas and executing optimized semantic vector similarity searches inside a combined relational database layer.
Document Persistence: Utilizing document-store databases to store unstructured execution payloads, dynamic agent states, and raw chat logs.
Hybrid RAG Pipelines: Combining relational/exact-match queries with vector-space searches for high-precision retrieval.
Production Reliability & Performance
Granular Tracing: Instrumenting end-to-end tracing to monitor agent execution steps, debug non-deterministic loops, and track token costs.
Automated Evals: Creating programmatic evaluation testing and scoring frameworks to benchmark agent accuracy before production deployment.
Core Backend Development: Writing clean, concurrent, asynchronous Python code to handle high-throughput foundation model APIs.
#J-18808-Ljbffr
Agentic Workflows & Memory Systems
Stateful Orchestration: Building and debugging production-grade, cyclic multi-agent workflows and state machines.
Context & Memory Engineering: Implementing multi-layered memory architectures including short-term memory for active turn execution, long-term memory for cross-session state persistence, and episodic summarization to manage token context windows.
Tool Call Management: Designing dependable function-calling patterns equipped with automated retry logic and self-correction handlers.
Data & Retrieval Infrastructure
Vector & Relational Storage: Managing relational metadata schemas and executing optimized semantic vector similarity searches inside a combined relational database layer.
Document Persistence: Utilizing document-store databases to store unstructured execution payloads, dynamic agent states, and raw chat logs.
Hybrid RAG Pipelines: Combining relational/exact-match queries with vector-space searches for high-precision retrieval.
Production Reliability & Performance
Granular Tracing: Instrumenting end-to-end tracing to monitor agent execution steps, debug non-deterministic loops, and track token costs.
Automated Evals: Creating programmatic evaluation testing and scoring frameworks to benchmark agent accuracy before production deployment.
Core Backend Development: Writing clean, concurrent, asynchronous Python code to handle high-throughput foundation model APIs.
#J-18808-Ljbffr