Software Architect, AI Systems and Platform

Okta•San Francisco, CA

3d•$274,000 - $376,200

About The Position

Okta is hiring two hands-on Software Architects for the team building Digital Workers, a suite of autonomous agents for identity operations and security. The team works directly with Okta co-founder and CEO Todd McKinnon and aims to bring the product to Okta’s more than 20,000 customers. The role combines startup-style ownership and pace with the resources of a public company. Architects will write critical code, resolve cross-team decisions, and guide engineers through production rollout. One role focuses on agent execution, harnesses, and evaluations; the other focuses on backend services, integrations, data, and platform reliability. The position involves defining shared contracts and architectural boundaries across model reasoning, agent harnesses, durable workflows, and authorization-enforcing services. Architects will evaluate and implement model and harness integrations, design secure and reliable backend services, shape transactional and analytics data architecture, and coordinate work across Product, Applied AI, and Platform engineering. The role requires depth in either AI systems architecture or platform architecture, alongside the ability to contribute across the full product. The technology stack includes Python FastAPI, React, TypeScript, AWS Amplify, Step Functions, Amazon Bedrock AgentCore, Lambda, SQS, EventBridge, API Gateway, RDS PostgreSQL, pgvector, S3, KMS, Secrets Manager, CloudWatch, and CloudTrail. OpenAI and Anthropic model API integrations and an analytics layer using ClickHouse or BigQuery are planned, and the architects will help select and implement them. Okta describes an immersive, in-person onboarding experience and a global community spanning more than 20 offices. The listed annual base salary range for candidates in the San Francisco Bay Area is $274,000–$376,200. The company also offers equity where applicable, bonus, and benefits including health, dental, and vision insurance, a 401(k), flexible spending account, and paid leave including PTO and parental leave, subject to applicable plans and policies.

Requirements

Record of leading architecture and delivery across multiple engineering teams, with recent production code personally implemented.
Deep knowledge of distributed services, API design, data consistency, cloud security, and production operations.
Experience simplifying systems while improving reliability, security, or delivery speed.
Ability to explain design tradeoffs, guide senior engineers, and remain accountable through implementation.
Depth in at least one role area—AI systems architecture or platform architecture—and ability to work across the full product.
For the AI systems focus: experience building and evaluating agent harnesses, context management, model integrations, and quality controls.
For the platform focus: deep production systems experience and practical knowledge of integrating model-driven workflows.

Nice To Haves

Experience with Claude Agent SDK, OpenAI Codex SDK, Strands Agents, LangChain Deep Agents, LangGraph, or a comparable custom runtime.
Experience with enterprise identity, security products, OAuth, OIDC, SAML, SCIM, or complex customer integrations.
Experience with ClickHouse, BigQuery, agent evaluations, post-training systems, or open-source infrastructure.
Experience building a new product within a larger enterprise while keeping technical decisions reversible.

Responsibilities

Define and implement shared contracts for tasks, events, permissions, approvals, evidence, and usage.
Set clear boundaries between model reasoning, agent harnesses, durable workflows, and services that enforce authorization.
Select an agent harness using representative identity tasks, failure tests, and operational evidence; keep model providers and harness components replaceable.
Design integrations with OpenAI and Anthropic APIs, accounting for data boundaries, latency, cost, version changes, and measured release quality.
Build reference services for tenant isolation, scoped tool execution, idempotency, long approval waits, and verified results.
Define transactional and analytics data architecture, separating PostgreSQL application state from the planned ClickHouse or BigQuery analytics layer.
Connect Product, Applied AI, and Platform engineers through clear ownership and practical APIs, and resolve dependencies that slow delivery.
Write critical implementations, review recovery and security, and guide technical leads through production rollout and customer expansion.
Lead agent execution, harnesses, and evaluations, or lead backend services, integrations, data, and platform reliability, depending on role focus.