Senior R&D Scientist

You'll be designing and validating new experiment templates, auditing team entries for compliance, and even using the API for basic data integration. Basically, you're a power user who helps define how we use it.

You'll be designing and analysing complex DoEs (Factorial, RSM), writing custom scripts (JSL) to automate analyses, and teaching advanced statistical concepts to junior team members. You're the go-to person for tricky stats.

You'll be writing custom functions and scripts from scratch for data cleaning, complex analysis, and advanced visualisation. You'll develop robust, reusable code that the whole team can use, and you'll be comfortable with version control (Git).

You'll be building new models from first principles, defining mesh and boundary conditions, and rigorously validating simulation results against experimental data. You'll interpret the models to guide experimental work.

You'll perform advanced structure/reaction searches, analyse citation networks, and conduct initial Freedom-to-Operate (FTO) searches to inform project direction and avoid infringing on existing IP.

You'll design and manage the information architecture for your projects or team, creating standardised reporting templates and dashboards. You're making sure our knowledge is organised and accessible.

• R&D Manager (for project direction and strategic alignment)
• Project Leads (for technical input and problem-solving)
• Engineering Team (for tech transfer and scale-up discussions)
• Manufacturing Team (for understanding process limitations and requirements)
• Quality Assurance (for method validation and compliance)
• Other Senior Scientists (for peer review and collaboration)

• Academic Collaborators (for specific research partnerships)
• Key Vendors (for new equipment or specialist materials)
• Contract Research Organisations (CROs) (for outsourced testing or analysis)

As data generation becomes more automated, the ability to build sophisticated predictive models and apply machine learning techniques to complex R&D problems will become standard. This means moving beyond basic regression to more advanced algorithms.

Supervised & Unsupervised Learning · Model Validation & Cross-Validation · Feature Engineering · Interpretability of ML Models

• This week: Explore the scikit-learn library in Python. Pick a simple dataset and try to build a basic classification or regression model.
• This month: Take an online course on applied machine learning for scientists or engineers. Focus on practical applications, not just theory.
• Month 2: Identify a dataset from a past project where ML could have provided deeper insights. Build a model and compare its performance to traditional statistical methods.
• Month 3: Present your findings and the potential benefits of ML to your R&D team, perhaps proposing a pilot project.

Quick win: Use a simple machine learning model (e.g., a random forest) to predict experimental outcomes on a small, well-understood dataset today.

The concept of creating a 'digital twin' – a virtual replica of a physical process or system – is gaining traction. For R&D, this means you'll be building more sophisticated simulation models that can predict performance under various conditions, optimising processes before we even run a physical experiment.

Multiphysics Modelling · Real-time Data Integration · Optimisation Algorithms · Uncertainty Quantification

• This week: Review existing COMSOL or MATLAB models in our library. Understand their inputs, outputs, and underlying physics.
• This month: Take an advanced course or tutorial on a specific aspect of simulation relevant to our work (e.g., finite element analysis, computational fluid dynamics).
• Month 2: Propose and build a small 'digital twin' for a simple lab process, linking it to real-time data if possible, to demonstrate its predictive power.
• Month 3: Collaborate with an engineering colleague to apply advanced simulation to a manufacturing challenge, showing how it can optimise a process.

Quick win: Identify a simple lab process that could benefit from a basic simulation model. Build a proof-of-concept in COMSOL or MATLAB to show its potential.