AI Systems Engineer

Job Overview:

You’ll join the team building the agentic systems at the core of Invictus — multi-agent orchestration, retrieval, and the production engineering that keeps it all reliable under real load. This is hands-on ownership role for an engineer who has shipped agentic systems into production and can take them from prototype to dependable, cost-aware infrastructure.

Requirements:

• Demonstrable experience building and operating agentic systems in production — multi-agent orchestration, tool routing, and human-in-the-loop — implemented with LangGraph and LangChain and extended well beyond framework defaults.
• 3–5 years of AI engineering experience overall, with a track record of shipping and maintaining systems that run in production at scale.
• A systematic approach to prompt engineering as a production discipline — design, versioning, and structured iteration backed by evals rather than ad-hoc tweaking.
• Strong debugging instincts in complex, non-deterministic systems running under real production load and traffic.
• Solid production engineering fundamentals — tested code, healthy code-review habits, and sound architecture trade-offs made with reliability, maintainability, and operational cost in mind.
• Working knowledge of RAG — hybrid search, re-ranking, query rewriting, and familiarity with retrieval evaluation metrics (recall@k, MRR, context precision/recall, RAGAS), including how these hold up against real-world data and live query patterns.
• Hands-on experience with CI/CD (GitHub Actions), Docker, and containerized deployment.

Nice To Have:

• Cost/latency optimization — semantic caching, model routing, batching, fallback strategies.
• Comfort building eval and observability tooling for non-deterministic systems, using LangSmith or Langfuse.
• Some experience with full-stack development.
• LLM gateway / provider-abstraction experience (LiteLLM, OpenRouter).
• Model serving or fine-tuning exposure (LoRA/QLoRA, vLLM, TGI).
• Experience with Azure or similar cloud platforms.