The opportunity
Secure Every Identity, from AI to Human Identity is the key to unlocking the potential of AI. Okta secures AI by building the trusted, neutral infrastructure that enables organizations to safely embrace this new era.
What you'll do
Implement intent-based enforcement to verify agent runtime requests match: their intended purpose, requiring key capabilities.
Apply LLM reasoning and prompt parsing to interpret prompts, tool payloads, and intent in real time.
Integrate low-latency inference or semantic evaluation engines directly into the API gateway request path.
Use embeddings, vector search, or zero-shot classification to score alignment: between agent intent and executed actions.
Design confidence-scored decision engines that feed semantic verification: results into policy frameworks (e.g., Cedar).
Establish evaluation benchmarks, prompt injection defenses, and guardrails to: prevent bypasses or false positives.
What they're looking for
- Architect scalable ML and Generative AI systems integrating retrieval, inference, and evaluation pipelines.
- Optimize prompting, context retrieval, and RAG workflows for accuracy,: safety, and efficiency in Claude-based systems.
- Build automated evaluation pipelines to measure model quality, correctness,: groundedness, and safety in production.
- Implement schema validation, structured output enforcement, and guardrails to: ensure reliable, compliant AI outputs.
