In my recent projects, I delved into the realm of Natural Language Processing (NLP) and developed advanced chatbots and semantic search applications. The primary focus was on creating intelligent systems capable of understanding and responding to user queries in a contextually relevant manner.I leveraged a variety of Language Models (LLMs) to enhance the conversational abilities of the chatbots. This involved experimenting with APIs or models such as GPT or LLAMA to find the most suitable fit for each specific project. The choice of LLMs played a crucial role in determining the quality and naturalness of the interactions.
I’ve had the privilege of crafting solutions tailored for the healthcare sector, specifically catering to hospital physicians and researchers providing robust tools for academical papers analysis and semantic search.
I also build apps for international commerce covering customs management and understanding the necessary paperwork.
I’ve ventured into the media industry building applications that have autonomously generated news articles in recent news, showcasing a fusion of technology and journalism. To enhance accuracy and veracity we linked to the URLs of the different articles where we extracted the facts.
Efficiency in information retrieval was a core concern. I implemented vector datastores to index and retrieve information quickly. This involved encoding documents and queries into high-dimensional vectors, allowing for fast and accurate semantic search capabilities.
To enrich the conversational experience, I integrated various sources of information . This includes but is not limited to structured databases, APIs like Jira, Confluence, NYTimes and even unstructured data from the web. Also non structured sources like academic papers PDFs or institutional publications like spanish BOE (in PDF). This approach enabled the chatbots to provide users with a broader range of relevant and up-to-date information. In my opinion, formatting of splitting on the best possible way is the key of success in those kind of projects
The implementations resulted in highly responsive and context-aware chatbots, providing users with more meaningful interactions. The semantic search applications showcased significant improvements in accuracy and speed, enhancing the overall user experience.
