Senior AI Engineer

The Incubate Team explores and implements cutting‑edge applications in Agentic AI and Machine Learning. The team is currently building a platform for the safe execution of AI agents: an environment where autonomous agents can run, call tools, and act on data with isolation, policy enforcement, observability, and human oversight built in.

As a Senior AI Engineer / Backend Developer you will own the backend architecture of this platform — the services, APIs, and runtime that make agent execution reliable, secure, and scalable.

Experience: 5+ years in backend software development

Location / Work model: Rome (Italy) / Hybrid

Mandate Skills

• Education: Bachelor’s or Master’s degree in Computer Science, Software Engineering, or a related field
• Backend development: 5+ years building and operating production backend services — REST/gRPC APIs, asynchronous processing, message queues, relational and NoSQL data stores
• Programming: Python (expert level); solid grounding in software design patterns, concurrency, and performance
• Agentic AI / GenAI integration: hands‑on experience building LLM‑powered services and agents — tool/function calling, orchestration (LangChain, LangGraph), RAG — and integrating commercial and open‑weight LLM APIs
• Security engineering: authentication/authorization, secrets management, sandboxing/isolation, audit logging, secure API design
• Scripting &; OS: Bash shell scripting on Linux/Unix

What you will do

• Design and build the backend of the agent execution platform: agent runtime and lifecycle services, tool/function‑calling gateway, sandboxed execution orchestration, policy and guardrail enforcement, and audit/observability pipelines.
• Develop robust, well‑documented APIs (REST/gRPC) and event‑driven components that let agents interact safely with enterprise systems, data sources, and MCP‑style tool servers.
• Engineer for production from day one: concurrency, queuing, retries with backoff, idempotency, rate limiting, multi‑tenancy, and LLM cost/usage controls.
• Integrate and operate LLMs and agent frameworks within the platform; build the abstractions that let the team swap models and providers without rework.
• Implement security controls across the stack — least privilege, isolation boundaries, input/output validation, secrets handling, prompt‑injection and data‑exfiltration defences — and wire safety/regression test harnesses into CI/CD.
• Own deployment and operations on Kubernetes: containerization, Helm/IaC, monitoring, alerting, performance tuning, and incident response.
• Provide technical leadership: lead architecture and code reviews, define backend and security standards, mentor engineers, and raise the engineering bar across the team.
• Partner with the AI Engineer / Data Scientist to productionize prototypes, expose evaluation infrastructure, and turn research results into shipped components.
• Collaborate with product, security, and platform teams; document designs through ADRs and communicate trade‑offs clearly.

What we are looking for:

• BE/BTech or above in Computer Science, Software Engineering, or related stream; 5+ years of backend product development, including 1–2+ years building LLM/GenAI‑powered services
• Expert‑level Python and strong software engineering fundamentals: clean architecture, automated testing, code review, version control, CI/CD
• Proven experience designing, shipping, and operating distributed backend systems at scale (APIs, async workers, message brokers such as Kafka/RabbitMQ, SQL/NoSQL databases, caching)
• Hands‑on experience integrating LLM APIs and agent frameworks (LangChain, LangGraph, LlamaIndex) and vector/retrieval stacks into production services
• Deep knowledge of Kubernetes and container‑based deployment; Docker proficiency
• Strong security mindset: sandboxing, isolation, AuthN/AuthZ (OAuth2/OIDC), secrets management, auditability, and awareness of the OWASP Top 10 for LLM Applications
• Bash shell scripting in Linux/Unix environments
• Ability to lead ambiguous, greenfield technical work from concept to production with limited supervision
• Excellent written and verbal communication skills; able to explain complex technical trade‑offs to mixed audiences

Nice to have

• Java, Go, or Rust for high‑performance or integration components
• Experience with container sandboxing and isolation technologies (gVisor, Firecracker, Kata Containers, WebAssembly runtimes)
• Familiarity with agent interoperability protocols (MCP, A2A) and tool/plugin ecosystems
• LLMOps tooling: model serving (vLLM, TGI), tracing/observability (OpenTelemetry, LangSmith), evaluation frameworks
• Cloud platform experience (AWS, Azure, GCP) and infrastructure‑as‑code