Machine learning scientist

Scindo is building the next generation of enzyme-powered chemistry by leveraging AI-powered enzyme discovery and design to reshape sustainable manufacturing through advanced biocatalysts. By creating unprecedented control over selectivity, our solutions offer innovative synthesis routes that reduce energy consumption, minimize waste, and decrease reliance on fossil feedstocks. Our enzymes enable direct conversion of natural, renewable, or upcycled materials into bioactive ingredients found in everyday products, such as cosmetics, personal care, and food. We are committed to transforming industrial chemistry for a sustainable future.

Role Description

You will contribute to the continued development of the machine learning models behind our enzyme function prediction and generative protein design. Working closely with our experimental team, you will help translate model outputs into testable designs, analyse results, and feed experimental data back into the next round of model development. You will also contribute to mining and curating our proprietary enzyme dataset to extract novel functional signals that drive the platform forward.

The role is based in our office and lab in central London.

Qualifications

• PhD (or equivalent) in statistics, machine learning, applied mathematics, computational biology/chemistry, or computer science.
• Demonstrated model development rather than model application: you have designed and trained novel architectures or training objectives, or substantially reworked published ones, and can derive and defend the objective of any model you work with.
• Strong foundations in probabilistic modelling and Bayesian inference, including Gaussian processes, variational inference, or uncertainty quantification.
• Experience designing and running active learning loops using Bayesian optimisation and probabilistic modelling, in settings where experimental throughput is the constraint.
• Strong programming skills in Python, at the level of writing custom layers, losses, and training loops in PyTorch or JAX.
• Ability to work independently while contributing effectively to a multidisciplinary team.
• Working knowledge of the architectures underpinning current protein and molecular ML, e.g. masked pretraining objectives, equivariance and how it is enforced, denoising diffusion and flow matching - at the level of having implemented or modified them.
• Experience with multi-task and multi-objective learning frameworks.
• Familiarity with probabilistic graphical models, hierarchical Bayesian models, or structured priors for scientific data.
• Experience with Monte Carlo methods, MCMC sampling, or stochastic variational inference.
• Familiarity with statistical learning theory, generalisation bounds, PAC learning, or information-theoretic approaches to model selection.
• Experience applying dimensionality reduction or latent variable models (VAEs, factor analysis, probabilistic PCA) to high-dimensional biological data.
• Familiarity with neural force fields or QM/ML hybrid approaches.
• HPC / GPU cluster experience, distributed training, performance optimisation.

What we offer

• The opportunity to work at the frontier of ML-driven enzyme design, with direct impact on real-world industrial chemistry.
• An exciting active feedback environment: model predictions are tested in-house by our wet lab, and you will work closely with experimentalists to design the experiments that make the models better.
• The chance to join a top interdisciplinary team and make a meaningful contribution to a rapidly developing platform.

#J-18808-Ljbffr