About Starya
Starya develops AI for complex operations. We connect artificial intelligence, software, and data to transform real business processes with security, reliability, and control.
Our platform, NebulaOS, is the foundation for building, integrating, and operating these solutions. Our work combines engineering close to customers, product development, and a foundation of data, infrastructure, and security.
The challenge
Turn customer needs into usable software, from investigating the process to learning in production. Combine programming, an understanding of data, and conversations with users to build integrations, configure platform capabilities, and verify results in the target system, handling exceptions that a demonstration with ideal data does not reveal.
What you will do
- Investigate tasks with customer users and specialists, mapping data sources, rules, and systems. Distinguish requests for a solution from the operational problem, and build verifiable examples, including situations where data is missing, inconsistent, or arrives after a decision.
- Code API- and event-based integrations and build workflows in NebulaOS. Model inputs, states, and contracts, apply validations and operational rules, and write tests that verify the effect of an action as well as the response presented by the AI.
- Implement authorized access and isolation between organizations in queries, connectors, and tool execution. Handle tenant context and permissions in code, test unauthorized access attempts, and produce useful diagnostics without exposing sensitive data or credentials.
- Handle failures and unknown outcomes: after a timeout, reconcile the AI workflow with the state in the ERP or CRM before repeating a write. Implement deduplication, bounded retries, and routing of pending work to avoid actions being duplicated or silently lost.
- Build paths for human intervention when data is incomplete, responses are inconclusive, or rules require exceptions. Bind approval to the action and parameters executed, require a new review when they change, and check the state before resuming the workflow.
- Deploy in increments agreed with the team and customer, verify results in production, and investigate differences between expected and observed behavior. Fix integrations and tests, document recovery, and bring recurring patterns to the product team with examples that support reuse.
What we look for
- Depth in programming with TypeScript and Node.js or equivalent languages and environments: implement services, organize testable code, understand asynchronous execution, and investigate errors. Ability to explain implementation choices and maintain components written by others.
- SQL knowledge to query and relate data, investigate inconsistencies, and confirm states. Understand granularity, transactions, and access boundaries to distinguish missing information from an incorrect query or transformation.
- Experience integrating APIs and events, including authentication, authorization, contract validation, and compatible evolution. Ability to reproduce communication failures and distinguish confirmed errors, pending operations, and still-unknown outcomes before repeating a call.
- Ability to discover rules with users and specialists, ask about exceptions, and turn that understanding into code and acceptance criteria. Communicate alternatives and limits without relying on technical vocabulary to explain their effect on the process.
- Practice in incremental delivery, testing, and debugging under operational conditions. Ability to evaluate AI responses and actions with representative examples, recognize model limits, and implement confirmation in the target system, recovery, and human review according to risk.
Additional experience
- Experience with AI applications that retrieve information or execute tools, including quality evaluation, reference data, and investigation of unexpected behavior.
- Experience integrating ERP, CRM, or healthcare, insurance, and business service processes, particularly state reconciliation and exception handling.
- Experience turning a specific integration into a reusable component, with documented contracts, behavioral tests, and maintenance agreed with the team receiving the capability.
Your impact
Your work makes it possible to check whether an action actually happened and recover tasks left pending. The proportion of confirmed actions, duplicates, unresolved work, and recovery effort, measured per operation, help determine what to fix, simplify, or expand.
Who you build with
We are building customer squads with one Client Engineering Lead and two FDEs. You work with the customer and lead throughout delivery, collaborating with SWE, PM, QA, research, design, data, SRE, and security as needed. Agree on interfaces and maintenance so that customer adaptations and NebulaOS development have clear owners.