Senior Software Engineer, Applied AI

Okta•San Francisco, CA

22h•$165,000 - $226,000

About The Position

The team that built Jeeva AI is developing Digital Workers, a suite of autonomous agents for identity operations and security at Okta. The team works directly with Okta’s co-founder and CEO and is building a product for Okta customers. In this role, you will build the harnesses and evaluations that make the agents reliable as they investigate risk, use tools, wait for approvals, and complete tasks across enterprise systems. You will own an agent capability or evaluation pipeline and deliver measurable improvements. The work includes developing agent execution capabilities, integrating and evaluating foundation-model APIs, building evaluation datasets and replay tests, analyzing production failures, and improving prompts, retrieval, tools, and model routing. You will help ensure actions are authorized, evidence-backed, safe, and reliable, including protecting tenant boundaries in retrieval and memory. The core stack is Python FastAPI, AWS Step Functions, Amazon Bedrock AgentCore, PostgreSQL on Amazon RDS, and pgvector. OpenAI and Anthropic APIs are part of the integration plan. The team plans to add an agent harness and a dedicated analytics layer using ClickHouse or BigQuery; the harness choice is open, and equivalent tools are considered.

Requirements

Strong Python engineering skills and experience shipping an LLM application that uses tools.
Hands-on experience building agent workflows or evaluation systems for tool-using LLM applications.
For harness work, relevant experience may include Claude Agent SDK, OpenAI Codex SDK, Strands Agents, LangChain Deep Agents, LangGraph, or a custom implementation.
Experience integrating model APIs, structured outputs, and asynchronous services; experience with OpenAI or Anthropic is especially relevant.
Depth in agent execution or evaluation systems, with working code and failure analysis that can be explained.
Ability to distinguish a plausible model response from a supported finding or verified action.

Nice To Haves

Open-source contributions, agent observability, security evaluations, or synthetic evaluation data.
Fine-tuning, preference data, post-training evaluation, or model distillation using approved datasets.
Experience with identity and security workflows.
Experience with tools equivalent to the planned agent harness or analytics layer.

Responsibilities

Build the agent harness for planning, tool use, memory, context compaction, interruption, and recovery during long-running work.
Integrate OpenAI and Anthropic model APIs and select models based on measured quality, latency, and cost, with clear failure handling.
Evaluate existing harnesses and build missing capabilities; keep the harness connected to AWS workflow execution and approval state.
Keep authorization and sensitive writes behind application policy checks, and require evidence of resulting state before marking work complete.
Build evaluation datasets and replay tests for valid findings, access decisions, app setup, and AI agent onboarding.
Measure false positives, missed risks, unsafe actions, completion rate, and cost; convert production failures into regression tests.
Test prompt injection, stale evidence, conflicting instructions, and unsafe tool use, while protecting tenant boundaries in retrieval and memory.
Improve prompts, retrieval, tools, and model routing.
Evaluate fine-tuning or other post-training methods when results justify them and data use is approved.
Own an agent capability or evaluation pipeline with measurable improvements.