AI/ML Consulting

Founded in 1980, Ashoka is the world's largest network of social entrepreneurs. Ashoka selects, supports and collaborates with 4000 social entrepreneurs (Ashoka Fellows) in over 90 countries, helping them scale their system changing solutions within every field, from education to health care, to environment, economic development and human rights. Ashoka was looking to create and implement an “Impact Strategy Bot” based on OpenAI’s GPT-3 model. The goal was to create an NLP model that generates an interesting strategy for addressing a social problem provided by a prompt from the user.

My task was to create a finetuned custom GPT-3 model for Ashoka to accomplish this goal. To achieve this I first analysed, cleaned, and prepared the more than 4,000 records that Ashoka provided as a potential training dataset. After diving deeper into the data I identified potential features that could be used for training GPT-3. I prepared the dataset accordingly and also engineered the appropriate prompts for the model to maximise the likelihood of the model creating interesting strategies. I then trained GPT-3 using OpenAI’s API. After a few iterations and testing the custom models together with Ashoka I was successful in creating an NLP model that would generate interesting strategies from a short prompt mentioning a social problem. The model will be implemented by Ashoka to be used by future social entrepreneurs within their platform.

Contract review is the process of thoroughly reading a contract to understand the rights and obligations of an individual or company signing it and assessing the associated impact. It is widely viewed as one of the most repetitive and most tedious jobs that junior law firm associates must perform. It is also expensive and an inefficient use of a legal professional’s skills. In this project I show how to set up a machine learning models to automate contract reviews.

As a volunteering data scientist at  DataKind UK I worked with Global Witness on a project to analyse the world’s first fully open register of the real owners of its companies. We were looking for mistakes and suspicious signs, while also comparing information in the register with other datasets. This provided us with an unprecedented overview of UK company ownership today and allowed us to identify loopholes, information gaps and suspicious activity.

​

Our analysis revealed that thousands of companies are filing highly suspicious entries or not complying with the rules – problems we never would have found were it not for the open data nature of the register.