Alin-Gabriel

In the corporate sector, text data flows swiftly, emanating from diverse sources in various formats, destined for multiple endpoints and actions. This continuous stream of data necessitates management through structuring, making it easy to interpret and analyse, and ultimately directing it to the appropriate user or applying the insights obtained. I utilised several large language models (LLMs) streamline and enhance this process:

• A question-answering LLM, enabling users to inquire about the content of each document within the flow.
• A summarization LLM, offering a concise overview of each document.
• An NLP model for sentiment analysis, determining the sentiment of each document.
• An NLP model that calculates an absolute similarity score among all documents.
• Clustering, which groups similar documents together and simultaneously identifies outlier documents.

In the dynamic realm of banking, ensuring effective anomaly detection is paramount. Utilising my expertise in state-of-the-art ML algorithms and advanced Deep Learning techniques, I tackled this imperative at Deutsche Bank, a globally recognized leader in financial services. The solution involved several critical stages:

• Identifying relevant data from various sources.
• Researching and testing optimal Machine Learning algorithms for tabular data analysis.
• Exploring and validating appropriate Deep Learning models for extracting insights from textual data (Natural Language Processing).
• Integrating the resulting models into the rigorous banking environment, complete with all necessary software engineering components: Python, Object-Oriented Programming, Algorithm Design and Complexity, Deployment, and more.

Market research is a domain teeming with diverse data types, including tabular, audio, and particularly text. This wealth of data can be utilised not only to enhance internal processes through automation but also to equip the end client with valuable new functionalities and use cases.

• The objective of this use case was to automate the labelling of open-ended questions from online market research surveys. Open-ended questions, which receive text responses, require further post-processing to extract useful information. I automated this process for one of the world's largest market research firms using a series of interconnected techniques:
• Utilising various forms of word embeddings, leveraging both Neural Network models and general techniques.
• Implementing clustering to group similar responses.
• Applying summarization and content extraction methods to distil and highlight key insights.