Principal AI/ML Engineer

This is a fully remote position, open to applicants in Belgium.

📋 Description

• Design and advance our multi-agent orchestration platform, currently leveraging Hermes / Multica.

• Create and implement voice AI pipelines that include STT, real-time TTS with streaming, VAD, and SIP/RTP telephony integration.

• Develop and maintain RAG pipelines featuring retrieval quality assessment, re-ranking, and hybrid searches across vector and keyword indexes.

• Fine-tune and assess LLMs for specific tasks within domains such as customer support, classification, and structured extraction.

• Oversee the AI observability stack, which includes tracing, instrumentation, cost tracking, and quality regression alerting.

• Establish and enforce guardrails for hallucination detection, PII redaction, output safety scanning, and rate-limiting in multi-tenant environments.

• Construct data ingestion, preprocessing, and feature pipelines that support model training and ongoing learning.

• Set architectural benchmarks for AI systems within the organization; conduct design reviews and oversee ADRs for key decisions.

• Guide ML engineers and applied scientists, enhancing the team's capabilities in production AI.

• Collaborate with external research partners and monitor emerging developments to identify valuable production signals.

⛳️ Requirements

• 8+ years of experience in ML Engineering, Applied AI, or Research Engineering, including at least 2 years in a lead or staff-level capacity.

• Profound, hands-on experience with LLMs in a production setting: fine-tuning, RLHF/DPO, prompt engineering, RAG, and tool usage.

• Proficient in Python and the essential ML stack, including PyTorch, Transformers (HuggingFace), and PEFT/LoRA.

• Practical experience with LLM inference serving technologies such as vLLM, TensorRT-LLM, or TGI in a latency-sensitive production context.

• Solid understanding of agentic frameworks, including multi-agent coordination, tool-call orchestration, context and memory management, and observability (Langfuse, Opik, or similar).

• Background in speech AI (ASR/TTS pipelines) or real-time audio systems is a significant advantage.

• Comprehensive knowledge of MLOps practices, including experiment tracking (MLflow/W&B), model registries, containerization (Docker/Kubernetes), and CI/CD for ML.

• Awareness of LLM-specific risks such as hallucination, prompt injection, data leakage, fairness, and privacy, along with strategies for mitigation in production.

• Excellent communication skills, enabling you to produce clear design documentation, lead effective architecture reviews, and articulate trade-offs to non-technical stakeholders.

• Experience with end-to-end voice pipelines: VAD → ASR → LLM → TTS → SIP/RTP telephony is a plus.

🏝️ Benefits

• We prioritize Diversity & Inclusion at our core.

• We have committed ESG efforts and ambitious sustainability objectives.