Software Engineer (AI Engineer)

At Allegro, we are redefining the e-commerce experience by building an Intelligent Shopping Assistant. We are creating a new conversational sales channel that understands not just what a user types, but why they are shopping. The Assistant is a trusted guide, supports, and inspires buyers with diverse needs to let them shop in ways that reflect their individual preferences, aimed at making Allegro the most intuitive and intelligent marketplace in Europe.

In the Product Intelligence team, our mission is to provide domain-specific grounding for our AI solutions. We measure our success with increased consumer confidence and trust.

• 2-5 years of commercial experience in AI/ML or Software Engineering with a proven track record of developing and deploying robust systems to production.
• Strong proficiency in Python and familiarity with standard development concepts: CI/CD pipelines, containerization, and basic observability.
• Familiarity with using AI-assisted development tools, like Cursor or Copilot.
• Understanding of distributed systems and microservices architecture.
• Experience in building and maintaining RESTful APIs in a cloud environment.
• Deep practical understanding of Prompt Engineering, Context Window Management, and interacting with Large Language Models as API endpoints.
• An analytical mindset to debug complex issues in non-deterministic AI systems.
• English proficiency (B2+ level).

Nice to have:

• Experience with developing MCP servers to expose internal data and functionalities to AI agents.
• Experience with LLM orchestration frameworks (ADK) and building multi-turn agents.
• Experience with Google Cloud, specifically with Vertex AI features.
• Experience with Kotlin for production-grade services development.
• Experience with Vector Databases or Knowledge Graphs.

Agentic Orchestration:

• Design, implement, and scale stateful, multi-turn workflows using frameworks like Agent Developer Kit, managing complex user intents.

Protocol Integration:

• Build production-grade APIs to expose AI capabilities, implement emerging interoperability standards, and secure transactional layers (MCP, A2A, ACP).

Evaluation-Driven Development:

• Systematically measure system performance, optimizing core metrics such as User Success Rate and Hallucination Rate.

Context Management:

• Collaborate with Data teams to build and maintain data-serving layers that enrich AI agents with real-time enterprise data.

System Robustness:

• Maintain high code quality through rigorous testing, active code reviews, and full-stack observability using tracing tools, like LangFuse.

Continuous Learning:

• Monitor the rapidly evolving landscape of Generative AI and Agentic Systems (e.g., autonomous agents, tool use) and propose how to apply these advancements to Allegro's ecosystem.