Moneybox has recently started investing in shoring up our intelligence-driven content curation in the app and building our personalisation programme to make the app more relevant to customers. As part of our efforts to ramp up personalisation, we’re looking at how we can better support customers by offering them intelligent solutions that can help navigate them and build their confidence at scale. As an ML engineer in Moneybox you would be working on a self-contained project to utilise LLM’s to provide support and recommendations to customers at critical points in their home buying and wealth building journeys. This is a groundbreaking exploratory project we’ll be running for multiple years as we work on the rest of our ML suite in the background, to test the limits of what we can do with unbounded and bounded models. Coming into Moneybox - you would be embedded in our decision science team - looking to build a suite of models that can utilise unstructured data from the customer in order to produce generated content relevant to an appropriate stage in a customer journey. While we might be exploring chatbots as a delivery mechanism for this - the models wouldn’t be limited to that use case as part of our product roadmap. While this role will entail a lot of research - we don’t expect you to design new models from scratch, but rather to use, tweak and deploy existing foundational models.
What you'll do
You will be solely responsible for researching an approach to generate content to guide customers that is relevant to them and their journey
Distilling open-source text generative models towards a lower parameter space to fit our use case at lower cost
Employing learning strategies to retrain the models to avoid common failure states in our domain
Employing RAG mechanisms to ground the models in relevant curated content
Tuning the models to achieve better performance
Supporting in evaluation mechanisms and ongoing monitoring to ensure models are performing within parameters when deployed at scale
Who you are
Have experience in using text generative models for a B2C scaled use case
Enjoy solving hard problems with uncertain outcomes
Enjoy optimising things and have an optimisation mindset when considering your problem solves
Are a systems thinker and enjoy figuring out a scalable solution that can fit an emerging system
Thrive in a fast-paced startup environment
Eager to learn new things and challenge your existing frameworks
Are not scared of ambiguity
Experience and skills - essential
2+ years of work experience as an ML researcher or engineer building, tuning and deploying scalable models in the text generation space OR equivalent academic experience with an industry application
Knowledge of applied machine learning, model tuning and model evaluation
Knowledge of the latest approaches in text generation, including latest LLM models
Experience in utilising open source LLM models and deploying them in cloud architectures
Experience with machine learning optimization, with demonstrated experience of navigating the cost/performance optimization through lowering the parameter space
Demonstrated experience of using good governance to track model performance, model versioning and model training
Experience and skills - not essential for the role, but will be counted as a plus
Experience in deploying using any of: Databricks, Azure or ML Flow
Research and building of foundational ML models, regardless of domain
Experience with using deep learning models combined with explainable models in a single decisioning system