AI Research Engineer (Speech & Conversational AI)

About Diabolocom Diabolocom enables organizations worldwide to build brand loyalty and drive sales growth through its cloud-based contact center platform. Our solution leverages AI-driven technologies to optimize customer experience across all communication channels. We provide seamless management of inbound and outbound interactions, premium voice quality, real-time analytics, and global scalability. Our platform integrates with leading CRMs and offers a wide range of APIs and services in multiple languages. With offices across Europe, the USA, Brazil, and the UAE, and more than 350 clients operating in over 60 countries, we are continuously growing and strengthening our engineering teams. About the role At Diabolocom, we build AI systems that operate on real-world customer conversations across voice and text channels. We are looking for a Research Engineer to bridge the gap between cutting-edge research and production-grade telephony AI. You will be responsible for end-to-end contribution across our research roadmap, with a focus on building production-ready systems for contact-center voice and text channels. What you will contribute to: Production Speech AI: Building high-performance pipelines for ASR, TTS, and Speech-to-Speech (S2S) workflows. Agentic LLM Engineering: Implementing fine-tuning, function/tool calling, and low-latency inference for real-time translation and agentic planning. Data-Centric Research: Designing scalable pipelines for synthetic data generation and automated quality evaluation. What we are looking for: ML/DL foundations: Solid grounding in modern ML/DL, with experience training models at scale. Speech & Audio: Expertise in S2S systems (Moshi, Qwen3 Omni), ASR pipelines (Whisper, Qwen3), TTS, and signal processing fundamentals. Software Engineering: Solid Python skills, with a focus on clean, reproducible code, testing, and Git workflows. Research Edge: A background in navigating ambiguity, reading/implementing SOTA papers, and a track record of contribution (papers, blogs, or open source). Valuable assets: Experience with speaker/voice anonymization or privacy-preserving techniques. Background in audio classification. Experience with real-time low-latency systems. Familiarity with cross-lingual speech models or bidirectional LMs. What we offer: Research with grounded impact: Bridge theoretical breakthroughs and production problems, solving them at scale. Contribute to open-source and in-house products: Work alongside a team of researchers and engineers to build our products and publish SOTA research. Recruitment Process 1. Introductory call with a Talent Acquisition Manager 2. Fit Interview with Théo, AI Researcher Lead 3. Research discussion with Théo, AI Researcher Lead 4. Paper presentation to the Research team