Ai Software Engineer
London, England, United Kingdom
Due to the advancements in AI, stakeholders have decided to research and implement machine learning (ML) solutions to improve the daily activities of an editorial team. This is aimed at creating a more productive workflow.My task involved studying and researching Natural Language Processing (NLP) techniques in the context of financial crime, that would ultimately benefit the project contributors and clients.During this process my responsibilities included:- Cleaning and analysing the entire corpus of articles.- Feature engineering based on existing categories, tags and keywords.- Training and evaluating machine learning models for text classification, summarisation and entity extraction using the new features.- Clustering and topic generation for identifying new themes.- Building a recommendation system using vector embeddings and cosine similarity.- Working closely with and guiding editors while collecting training data.- Prompt engineering utilizing zero shot, few shot and chain of thought (CoT) techniques with large language models (LLM).- Setting up a Python web server for hosting proprietary models and implementations.- Planning and creating API routes for communication between back-end systems.Technologies and tools used:- OpenAI GPT 3 / 3.5 / 4 and function calling API- Hugging Face- Python- Flask- Pinecone- LlamaIndex- Transformers- SpaCy- Sklearn