AI Engineer

Ruya AI•Massachusetts (US) - Onsite, MA

2d•Hybrid

About The Position

Ruya AI is a venture-backed defense-tech company building AI-native intelligent software for sovereign institutions and their affiliated organizations operating in mission-critical, high-stakes environments. Headquartered in Massachusetts (US), with offices across the U.S., Europe, and the Middle East, we are expanding our engineering team with people who demonstrate strong ownership, sound judgment, and the ability to turn complex requirements into reliable products. The Role Quant is the team responsible for our multi-modal intelligence agent. Our software is built to surface hidden patterns across disparate data sources, and generate predictive outcomes for the operators who rely on it. As part of this team, you'll design and build the agent systems, tool integrations, and evaluation pipelines behind those capabilities. You'll own the orchestration layer that ties those systems together, converting complex operator requirements into secure, reliable, and measurable AI capabilities. You'll work hands-on, building and operating these systems in close partnership with the broader engineering team. If you're excited about building autonomous agent systems that operate reliably in mission-critical environments, this is the right role for you.

Requirements

Strong production Python and asynchronous programming experience.
Experience building LLM applications with tool use and multi-step reasoning.
Knowledge of structured output validation and non-deterministic system testing.
Practical understanding of prompt injection and sensitive-data risks.
Ability to evaluate AI behavior empirically rather than relying on demonstrations.

Nice To Haves

Experience with MCP, Langfuse, LiteLLM, Anthropic-compatible APIs, agent memory, CopilotKit, geospatial intelligence, or entity graphs. Equivalent model and orchestration platforms are welcome.
Fine-tuning and pre-training of LLM models
Post-training
Agent and memory graph engineering
RAG and knowledge systems

Responsibilities

Build asynchronous Python/TypeScript services with FastAPI, Pydantic, and asyncio.
Develop agent orchestration, subagent delegation, tool calling, and structured outputs.
Design MCP tools and AG-UI/SSE integrations.
Improve prompt, context, memory, and token-management strategies.
Build safeguards against prompt injection, data leakage, and unsafe tool use.
Create behavioral evals and regression datasets using Langfuse.
Trace model calls and tool execution with Langfuse, Sentry, and OpenTelemetry.
Diagnose failures spanning model behavior, tools, data sources, and distributed services.