Senior Software Engineer

## About the role - Build core backend services in Python (FastAPI, asyncio, SQLAlchemy async) that power our attack-surface analysis engine and agent orchestration - Design and evolve the graph data layer (Neo4j + Cypher) that models enterprise security knowledge — assets, identities, configurations, attack paths - Integrate and extend our AI agent system built on Anthropic's Claude (Pydantic AI, Anthropic SDK) — agents that classify intent, plan response, analyze security graph state, and reason about vulnerabilities - Deploy and operate on Kubernetes (Azure AKS) with Docker, Helm, GitHub Actions — production observability via OpenTelemetry - Make architectural calls in collaboration with a small team (today: ~7 engineers across Bucharest, Seattle, and the Bay Area); your code will be in production within days, not quarters ## What we look for - 8+ years building production backend systems, ideally at a B2B SaaS company shipping to enterprise customers - Strong Python: async (FastAPI / asyncio / SQLAlchemy 2.0), production-grade error handling, performance debugging - Graph database experience (Neo4j, Memgraph, ArangoDB, JanusGraph, or similar) — or compelling track record of picking up novel data layers fast. Cypher fluency or willingness to ramp in 4–6 weeks. This is the single highest-leverage signal in your profile - Postgres + relational schema design — you can model a domain in either world and know when to use which - Containers + Kubernetes basics — you can read a Helm chart, debug a pod, and write a reasonable Dockerfile - Async writing in English — you can think in writing, document decisions clearly, and communicate across timezones without losing nuance ## Bonus points - LLM SDK integration experience (Anthropic, OpenAI, or similar) — bonus for agent orchestration patterns (Pydantic AI, LangGraph, etc.) - Cybersecurity domain background (threat modeling, MITRE ATT&CK, vulnerability management, identity, cloud security posture) - Distributed workflow engines (Restate, Temporal, Cadence) or stream processing (Flink, Kafka Streams) - Azure-specific experience (AKS, ACR, ServiceBus, Workload Identity Federation, AI Foundry) — AWS knowledge transfers fine - Java 21 (we have one JVM module for Azure event ingestion — not your primary work, but bonus if you can read it)