Workflow Engineer - Orkes; m​/f​/d

Role Overview

We are seeking a Workflow Engineer to design and implement the business process orchestration layer of an agentic AI platform at a leading UAE financial institution, using Orkes Conductor. You will translate complex banking process requirements — including clearing SLAs, HITL wait policies, compensating transaction rules, and retry strategies — into durable Orkes workflow definitions. Serving as the primary integrator between the AI Gateway, Lang Graph agents, core banking systems, and the Ops reviewer interface, you will ensure that AI cognition is woven into coherent, observable, and resilient banking workflows.

• Orkes workflow design: Design and implement Orkes Conductor workflow definitions for all AWA use cases — defining task sequences, sub-workflow compositions, HITL Wait tasks, fork/join parallel execution patterns, and SLA-governed task timeouts.
• HITL integration: Configure Orkes Human Task definitions for reviewer queue routing; implement HITL resume logic that correctly restores Lang Graph checkpoint state after reviewer decisions (Approve/Edit/Reject).
• Compensating transactions: Design and test compensating workflows triggered on downstream failures; ensure idempotency of all compensation steps through job .
• Retry and resilience: Configure Orkes task retry policies (fixed, exponential backoff with jitter), circuit breakers, and dead-letter queue handling for all task failure modes.
• Orkes and Lang Graph integration: Design the interface between Orkes tasks and Lang Graph agent invocations — passing job IDs through task inputs/outputs, managing Lang Graph execution as Orkes worker tasks, and surfacing Lang Graph errors correctly.
• Job n: Ensure unique job identifiers are correctly passed through every Orkes task input and output; validate via integration tests.
• Workflow observability: Configure Orkes workflow metrics export to Azure Monitor; build KQL queries for task failure rates, workflow duration distribution, HITL queue depth, and retry rates by use case.
• New use case onboarding: Lead workflow design for each new use case — define workflow skeletons, agree task SLAs with the business, design HITL routing policies, and build workflow test suites.

• 5+ years in workflow orchestration engineering using Orkes Conductor, Apache Conductor, Temporal, or Apache Airflow.
• Strong understanding of distributed systems concepts: idempotency, at-least-once delivery, eventual consistency, and saga patterns.
• Experience designing long-running durable workflows with human approval steps.
• Python or Java for Orkes worker implementation.
• REST API design: solid understanding of webhook patterns for async workflow resume.
• JSON/YAML workflow definition authoring and schema validation.
• Bachelor's degree in Computer Science, Software Engineering, or equivalent.

• Direct experience with Orkes Conductor — familiarity with Orkes Cloud, worker SDK, and Human Task definitions.
• Lang Graph or Lang Chain awareness — understanding of agentic execution models that Orkes orchestrates.
• Kafka consumer group management for event-driven workflow triggers.
• Azure Monitor and Log Analytics for Orkes metric export and alerting.
• Banking process knowledge: STP, clearing windows, and exception handling flows.
• Demonstrable experience with production workflow orchestration systems managing thousands of daily workflow instances.

We encourage applications from all qualified candidates, regardless of race, gender, disability, or any other characteristic that makes them unique.