Machine Learning Engineer - AI for 2D & 3D data

About Dynamic Optics

Dynamic Optics is a CNR-IFN spin-off and technology leader in adaptive optics, specialising in measuring, analysing, and correcting the optical behaviour of real-world lenses. Based in Padova and backed by Officina Stellare SpA — an Italian leader in advanced optical instrumentation for space — Dynamic Optics combines academic novelty with industrial-grade engineering and real-world deployment.

We are at the centre of a pioneering deep-tech project, co-funded by the European Union, at the intersection of cinematography, optical engineering, imaging science, and machine learning. Its mission is to digitise the optical behaviour of real professional cinema lenses, enabling precise measurement, simulation, and creative manipulation of characteristics such as distortion, aberration, flare, bokeh, and depth of field. By combining advanced optical engineering, computational imaging, and AI-driven modelling, the project creates high-accuracy digital lens twins — bridging physical cinematography with digital image generation and unlocking new capabilities for VFX, virtual production, CGI rendering, and generative AI.

With EU funding now secured, we are moving from a funded research action towards commercial deployment, and are building the team that will take this technology to the global audiovisual market.

The Role

We are seeking a Machine Learning Research Scientist to drive the project’s foundational research into physics-guided, optical-system machine learning.

We are looking for a Machine Learning Research Scientist with a strong background in Computer Vision, 3D data processing, and AI for scientific imaging. The ideal candidate has experience developing deep learning models for image understanding, 3D representations, or computational imaging, and is interested in combining data-driven approaches with physics-based optical modelling.

Your work will focus on developing new ML formulations, models, and experimental approaches for modelling optical behaviour — including PSF fields, aberrations, flare and ghosting behaviour, wave-optics approximations, and other lens-driven effects.

You will work at the research core of the optical modelling pipeline, designing algorithms that push the boundaries of optical ML and computational imaging.

Key Responsibilities

Foundational research and model development:

● Investigate and develop new cutting-edge ML models for optical behaviour estimation and simulation.

● Design deep learning models for computer vision and 3D optical data.

● Develop AI methods for learning from high-dimensional imaging and 3D measurement data.

● Develop novel techniques for modelling PSFs, vignetting, Zernike fields, flare and ghost behaviour, scatter, and chromatic effects.

Experimentation and prototyping:

● Build research prototypes to test new algorithms, architectures, and loss functions.

● Evaluate performance using scientific benchmarks, simulation outputs, and rendering-based validation.

● Maintain rigorous experiment tracking and reproducibility.

Data and measurement interpretation:

● Analyse complex optical datasets (PSF stacks, Zernike maps, spectral captures, flare and ghost imagery).

● Build data pipelines tailored to research experimentation.

Collaboration and scientific contribution:

● Work closely with rendering, simulation, and software teams to transition research into production ML systems.

● Develop technical documentation, research notes, and internal publications.

● Keep abreast of research in computational imaging, scientific ML, neural optics, and physics-guided modelling, and contribute to internal R&D strategy and long-term modelling direction.

Skills and Experience

Required:

● PhD (or MSc with equivalent experience) in Computer Vision, Machine Learning, Computational Imaging, 3D Vision, Computer Graphics, or a closely related field.

● Strong knowledge of deep learning architectures for image analysis and visual data (CNNs, Vision Transformers, diffusion models or related approaches).

● Proven experience developing deep learning models for computer vision applications.

● Experience working with 3D data (point clouds, meshes, volumetric data, multi-view imagery, neural representations, or scientific imaging).

● Strong mathematical grounding in optimisation, linear algebra, and Fourier optics.

● Proficiency with PyTorch and OpenCV.

● Strong mathematical grounding: Fourier optics, PDEs, optimisation.

● Experience working with scientific or imaging data.

● Fluent English, written and spoken (B2 or above).

Bonus:

● Publications in ML, computational imaging, optics, or graphics.

● Experience with 3D computer vision, neural rendering, NeRFs, Gaussian Splatting, differentiable rendering, or geometric deep learning.

● Experience modelling PSFs, wave-optics, or image-formation physics.

● GPU compute or high-performance ML skills.

● Prior experience in optics, geometric optics and ray propagation.

● Experience collaborating with teams in R&D-heavy or deep-tech environments.

What We Offer

● A research-driven role at the heart of the project’s modelling science.

● Direct collaboration with optical scientists, ML researchers, and rendering engineers.

● A fully remote, flexible working environment across Europe or the UK.

● Competitive compensation commensurate with experience 30-50k€.