Skip to content
Intesa Sanpaolo loses €95M to a cloned voice.
All open roles
EngineeringFull timeHybrid, MENA

Senior AI engineer (Multimodal classification & detection)

At Raid AI, we are building a world class detection engine from scratch to identify deepfakes across every medium: audio, video, images, and documents. We aren't looking for someone to maintain existing systems; we need an architect who can look at a research paper on Sunday and have a custom PyTorch classification model running by Thursday.

Key responsibilities

  • Build from zero: architect and implement multimodal classification frameworks to distinguish between real and AI generated content (including forged signatures and stamps).
  • LMM fine tuning: leverage and fine tune Large Multimodal Models (LMMs) to detect sophisticated synthetic patterns across audio, visual, and textual data — specifically targeting cross-modal inconsistencies and generative artifacts.
  • Optimisation & quantization: own the model lifecycle by implementing quantization (bitsandbytes, GPTQ) and optimisation techniques to ensure high speed inference.
  • Rapid research implementation: translate the latest state-of-the-art (SOTA) research into production-ready code.
  • End to end pipeline: manage the full pipeline — curating data and training on AWS, then optimising for deployment on GCP.

Qualifications

  • Experience: 4+ years of professional experience in Machine Learning / AI, with a proven track record of shipping models to production.
  • PyTorch native: expert in the PyTorch ecosystem.
  • Multimodal mastery: deep understanding of signal classification across audio, computer vision, and NLP.
  • LMM & optimisation expert: proven ability to fine tune and deploy Large Multimodal Models (LMMs) and Vision language Models (VLMs) for complex classification tasks.
  • Infrastructure: comfortable handling high-compute training on AWS and scalable deployment on GCP.
  • The builder mindset: you enjoy the "blank page" phase of a project and can navigate ambiguity to build robust systems.

Apply for this role

PDF only, max 10 MB. We use your information solely to evaluate this application.