Senior AI Engineer (Voice / Multi-Agent / Low-Latency Systems)

Location: Remote-first, Europe (Spain or UK preferred)

Office: Barcelona available

Engagement: Full-time, long-term

Employment: In-house

About the Company

Our client is a European health technology company building an AI-driven virtual assistant for clinicians — a real-time system that helps healthcare professionals perform tasks faster and more accurately.

Their mission is to build intelligent, low-latency AI agents that communicate naturally through text and voice, supporting healthcare providers, payers, and pharmaceutical companies.

Core Stack: Python, FastAPI, Docker, Kubernetes, Argo CD, gRPC, Redis, Postgres, WebRTC

Focus Areas: Real-time AI agents, voice systems, event-driven architectures, multi-agent orchestration

Who We’re Looking For

This is a Senior-level (6–7+ years) AI Engineer role.

We’re looking for someone who is highly experienced in building AI/ML systems - not just experimenting with the latest APIs, but shipping real production systems.

You are a software-first AI engineer:

• You understand architectures, reliability, observability, SLAs, and production constraints - not just prompts.

Mandatory background:

• Strong AI/ML expertise + solid software engineering fundamentals
• Experience that predates the LLM boom, showing a foundational understanding of ML/Software Engineering beyond prompt engineering.
• Proven experience with production AI/ML systems
• Previous startup experience (required)

Deep expertise in at least one:

• Voice AI systems
• Multi-agent architectures
• Low-latency software systems

What You’ll Own

You will take ownership of critical AI infrastructure components, including:

AI Brain Service (End-to-End)

• Architecture design, SLAs, latency budgets
• Failure modes & reliability engineering
• Production rollouts and system evolution

Real-Time & Voice Systems

• Streaming text/voice
• VAD, barge-in, turn-taking, interruption handling
• WebRTC, SIP, LiveKit (strong plus)
• Latency optimization under real-world constraints

Multi-Agent Orchestration

• Planner–executor–critic patterns
• Role routing & shared memory
• Tool routers & coordination protocols
• Evaluation-driven iteration

RAG & Retrieval Engineering

• Hybrid retrieval & re-ranking
• Query rewriting, compression, caching
• Freshness & grounding
• Faithfulness and relevancy evaluation

Evaluation & Observability

• Pre-call validation & safety enforcement
• In-call tracing (prompts, tools, tokens, latency, cost)
• Post-call automated evals (hallucination, safety, regressions)
• OpenTelemetry instrumentation
• Shadow evals & drift detection

Minimum Qualifications

• 6–7+ years of ML/backend engineering in product environments
• Strong Python expertise (FastAPI, asyncio, pydantic)
• Experience shipping production AI/ML systems
• Experience in startup environments
• Real-time system experience
• Strong understanding of architecture, observability & reliability

Nice to Have

• DSPy, MiPRO, GEPA
• LLM evaluation frameworks / LLM-as-judge setups
• WebRTC/SRTP, SIP, TURN/SFU tuning
• GCP (Cloud Run, GKE, Pub/Sub, Vertex AI)
• Healthcare domain experience

Example Problems You’ll Solve

• Push median voice round-trip below 2 seconds while preserving turn-taking.
• Implement OTEL-first tracing for agent graphs with automated evaluation triggers.
• Improve RAG with hybrid retrieval + measurable faithfulness gains.
• Turn EHR integrations into reliable LLM tools.

Working Style

• Remote-first (EU time zones preferred)
• Barcelona office available
• Direct collaboration with Head of AI and Engineering
• Fast iteration, high ownership
• Strong architectural influence