Principal Software Engineer, Applied AI
Core
Building autonomous agents (Digital Workers) for identity operations and security, integrating LLMs to execute tasks across enterprise systems.
Role type
Principal Software Engineer (Applied AI / Agent Systems)
Builds
Autonomous agent harnesses, evaluation systems, and tool-use workflows for identity security.
Domain
Identity & Access Management (IAM) + Generative AI
Deliverable
production ML models
Required skills
Python, LLM application development, agent workflow orchestration, model API integration (OpenAI/Anthropic), evaluation system design, failure analysis, prompt engineering, security policy implementation
Preferred skills
Agent SDKs (LangChain, LangGraph, Claude Agent SDK), fine-tuning, post-training evaluation, ClickHouse/BigQuery, open-source contributions
Technologies
Python, FastAPI, AWS Step Functions, Amazon Bedrock AgentCore, PostgreSQL, pgvector, OpenAI API, Anthropic API, ClickHouse, BigQuery
Responsibilities
Build agent harnesses for planning, tool use, memory, and recovery; Integrate and select LLM models based on quality/latency/cost; Evaluate harnesses and build missing pieces; Implement authorization checks and evidence validation; Create evaluation datasets and replay tests; Measure production metrics (false positives, completion rate, cost); Test for prompt injection and unsafe tool use; Improve prompts and model routing strategies.
Seniority
Principal, hands-on IC with product direction
