Track record
Generative AI Solutions
Using LLMs and state-of-the-art techniques such as multi-agent systems and advanced prompt engineering to create generative AI solutions.
Previous projects:- AiSentr: delivered prototypes using Azure GPTs, RAG and a multi-agent framework, to help close an important partnership for the business.
- Automatic Classifiers for Peace: delivered an LLM-powered classifier to aid peacebuilders in finding insights from social media data. Open source repository and community here.
End-to-End AI Platforms
Building and delivering complete AI platforms that cover the full spectrum from data collection and processing, start-of-the-art ML model development, to data visualization, derivation of insights, and automated decision-making.
Previous projects:- Phoenix: a collaboration between Build Up and datavaluepeople, has helped 100+ peacebuilding organisations to do large scale data-driven evidence-based initiative development through social media analysis, read more in our case study.
- LoveHolidays: established the core mono-repo, python libraries, infrastructure, workflows, for the data science team, powering 10s of effective data science projects across the business.
Advanced ML Modeling
Building state-of-the-art models that help to make novel and strategically important decisions.
Previous projects:- For a top 10 worldwide pharma company, we developed a batch manufacturing optimisation model, able to predict end-product quality of the batch production process, enabling reactive process corrections and optimisations.
- LoveHolidays: developed a pricing and ad-auction bidding model making millions of decisions daily, leveraging hierarchical bayesian modelling to get accurate predictions in a massively high dimensional, thus very data sparse, prediction space.