About Gladia

Founded in 2022 by Jean-Louis Queguiner (ex-OVHCloud) and Jonathan Soto (ex-MIT/Sigfox), Gladia builds best-in-class speech AI tools that empower businesses to deliver faster, more accurate, and innovative communication solutions worldwide.

Headquartered in vibrant tech hubs like Paris and New York City, Gladia is leading the charge in speech AI innovation. In under two years, we’ve grown exponentially, now serving over 150,000 users and 700 enterprise clients—including industry leaders like Attention, Circleback, Method Financial, and VEED.IO.

Our API supports advanced speech recognition and analysis in over 100 languages, setting a new standard for speed and accuracy across customer support solutions, voice agents, meeting assistants, and more.

Backed by world-class investors like Sequoia Capital, New Wave, and XAnge, we recently raised $16M in Series A funding—bringing our total funding to $20.3M. This investment powers our mission to build the ultimate AI audio infrastructure for leading platforms across the globe.

We are hiring a Staff Applied Audio Research Engineer to lead the development of next-generation audio models and processing pipelines at Gladia — from architecture exploration and model training to large-scale deployment. This role sits at the intersection of applied research, model engineering, and production-grade ML. It requires deep audio expertise, an experimental mindset, and hands-on experience bringing advanced models into real-world production environments.

The ideal candidate has already trained and deployed neural audio models at scale and is capable of guiding small teams or acting as a tech lead for strategic projects. Staying ahead of the state of the art, structuring technical watch, and bringing research to life are core to this role.

Role

• Research, train, and optimize models for ASR, segmentation, diarization, VAD, enhancement, etc.

• Explore and benchmark architectures such as Seq2Seq, RNNT, Transducers, Conformers, with strong attention to latency, memory, and streaming constraints

• Drive the end-to-end lifecycle of audio models: data pipelines, training, evaluation, deployment

• Drive improvements in latency, throughput, and accuracy

• Improve runtime performance through quantization, pruning, fusion, batching strategies, etc.

• Collaborate with Infra, Dev and Product teams to ship robust, scalable audio ML features and influence architectural decisions across the stack, from model design to deployment.

• Maintain a structured technical watch, benchmark state-of-the-art methods, identify strategic opportunities

• Leverage signal processing knowledge to improve feature extraction and system robustness

• Mentor other researchers or engineers and contribute to Gladia’s R&D direction

Bonus Points

• Contributions to open-source inference projects:

  • vLLM

  • Text Generation Inference (TGI)

  • Triton Inference Server

Profile

• 8 - 10+ years of experience in audio ML, applied research, or deep learning engineering with a focus on high-performance inference systems

• Strong expertise in Python and a low-level language such as C++, Rust, or Go.

• Proven experience training and deploying deep learning models for audio or speech and deep understanding of system-level performance optimization.

• Excellent command of audio-specific architectures: RNNT, Seq2Seq, Conformers, Transducers…

• Experience with ASR toolkits (e.g., Kaldi, ESPnet, Fairseq, OpenAI Whisper).

• Hands-on experience with inference frameworks like ONNX, TorchScript, TensorRT, or Triton Inference Server

• Hands-on experience with training at scale (multi-GPU, DDP, mixed precision, large datasets)

• Experience optimizing inference performance in production environments (latency, memory, CPU/GPU targets) and strong grasp of GPU architecture, CUDA programming, and latency/memory optimization.

• Experience in open-source-first environments.

• Strong background in digital signal and familiarity with model compression techniques and batching strategies.

• Comfortable working in a collaborative, fast-paced startup environment.

What Gladia offers

• Full remote policy with team gathering in Paris every 1.5 months or work from our amazing offices & rooftop in the heart of Paris (Sentier)

• Lunch vouchers (approx. 200€/month)

• An allowance of 360 euros/year for your sports activities

• An allowance of 300 euros for the adaptation of your workstation

• A health insurance (100% coverage) and a pension contract with Alan Blue for you & your family

• Unlimited vacation policy

At Gladia, we thrive on creativity, collaboration, and a shared passion for pushing the boundaries. Our team is made up of brilliant AI minds, all working together to deliver solutions that make a real-world impact. Whether you're looking to solve complex challenges, innovate in a fast-paced environment, or be part of a global movement transforming communication, Gladia is the place for you !