Subject: [AUDITORY] Research positions (audio/AI, 6-month contracts); Queen Mary University of London From: Iran Roman <iran@xxxxxxxx> Date: Wed, 24 Jun 2026 20:19:27 +0100Hi all, Two Research Assistant positions are open in my group at Queen Mary = University of London (QMUL). Both are 6-month contracts (part time or = full time), starting as soon as possible, at QMUL Grade 4.25 or higher = depending on experience. You must have the right to work in the UK. = Please apply to ONE position only. To apply: email i.roman@xxxxxxxx including a short note on your = interest and fit, plus a 1-page CV. That's it (feel free to email me = first if you have questions). Applications will be reviewed on a rolling basis until the posts are = filled. =E2=80=94 1. ACOUSTIC IMAGING & EDGE DEPLOYMENT We have self-supervised models that reconstruct high-resolution acoustic = spatial images from sparse microphone arrays, able to run in real time. = The work now is deploying them onto constrained, wearable hardware and = validating that they hold up in real-world, noisy conditions. Skills required: acoustics, multichannel array processing, embedded = systems deployment. Deliverable: a working real-time system. The successful candidate will work directly with the project's industry = partner, an undisclosed multinational company at the forefront of = artificial intelligence and immersive technologies. =E2=80=94 2. MULTIMODAL MACHINE PERCEPTION & GENERATIVE AI The core bottleneck in spatial computing is real-world multimodal data. = Existing datasets are either spatially blurred, too narrow in scope, or = synthetic. We're capturing our own: 360=C2=B0 video, 64-channel = full-sphere audio, egocentric perspective with head-tracked signals, and = sub-millimetre source tracking, across real kitchens, lounges, and = meeting rooms. You'll use this to train latent diffusion models and multimodal = transformers for scene synthesis, enhancement, and object segmentation. = Your models will anchor generative multimodal AI workshops and = challenges at international venues. Skills required: multimodal signal processing, generative modelling = (diffusion models, transformers), data capture and pipeline engineering. = Prior experience with multimodal signal processing is a real advantage. Best, Iran=