Senior AI Native Engineering
Віддалений* формат співпраці з території - Україна
*Ви можете працювати віддалено з території країни (або країн), для яких відкрита ця позиція.
AI Native Engineering
We are seeking a Senior AI-Native Engineer to become part of our AI-native team. This position focuses on architecting AI systems that fellow engineers and customers rely on, deploying them into production, and evaluating their effectiveness. The role follows an AI-native methodology: specifications come first, testing comes first, and coding agents handle implementation while under review.
Чим ви будете займатися у цій ролі
- Design and operate production LLM systems, including multi-stage pipelines or agents with a curated tool set, output schemas, validation, retries, and failure recovery
- Build and maintain evaluation frameworks: golden datasets, regression evals for every prompt change, LLM-as-judge with spot checks, and production telemetry
- Determine where tasks should be fully autonomous, where AI should assist, and where human approval is required
- Execute work using a spec-first and test-first approach alongside coding agents, while maintaining deliberate and token-aware project context files (AGENTS.md, CLAUDE.md, steering files)
- Integrate AI steps into CI/CD pipelines (review gates, test generation, quality hooks) and fine-tune them to reduce false positives
- Forecast AI costs before development and track them afterward, applying cost-control levers such as model routing, context trimming, caching, batching, and limits
- Create and maintain reusable skills, prompts, tool schemas, eval sets, and MCP servers
- Identify AI failures early and resolve them through structural fixes, while supporting new team members in adopting the team's AI workflow
Навички
- At least one AI workflow or system currently running in production or CI with minimal daily oversight, which you can describe in detail: the trigger, the users, your specific contribution, and how success is measured
- Experience designing multi-step LLM or agent systems, including tool restriction, output schemas, validation, fallbacks, and latency trade-offs
- An evaluation methodology that you've either built from scratch or enhanced
- A past experience where you corrected a flawed output through a structural fix rather than simply rerunning the process
- Understanding of what drives costs in your own AI work, along with hands-on use of at least one cost-reduction lever with measurable results
- A minimum of 5 years in professional software engineering with direct production ownership
- At least 2 years of hands-on experience building with LLMs or agents in real-world projects
- Strong communication skills, such that a colleague could recreate your work based solely on your explanation
Буде перевагою
- Experience with multi-agent systems in a production environment
- Background in brownfield discovery within a legacy codebase
- Experience embedding AI steps into CI/CD with measurable results (PR cycle time, defect density, AI acceptance rate)
- Familiarity with multiple model providers or platforms, along with the rationale for choosing each
- Experience with tracing and observability tools for LLM systems
- Background working in security-sensitive environments