Sign In

Andrea D'Agostino

Data Scientist. I manage and develop AI enterprise solutions in Italy and abroad. Traditional machine learning and LLMs. Data science and machine learning writer.

Andrea D'Agostino is a seasoned professional with 7 years of experience specializing in applied data science for digital marketing and data mining. Currently, Andrea's focus lies in the OSINT domain, where they oversee teams and lead machine learning projects and operations.

Andrea holds a Master of Science degree in Cognitive Science with a concentration in Cognitive Neuroscience from the Università di Trento. They also pursued a Bachelor of Arts degree in Cognitive Psychology Sciences and Cognitive Ergonomics at the Università degli Studi 'Suor Orsola Benincasa'.

Throughout their career, Andrea has held prominent positions in various organizations. They served as the Lead Data Scientist at DigiTrails, founded their own venture DagTech, worked as a Back End Developer at Braiins, contributed as a Data Scientist at Pro Web Consulting within Cerved Marketing Intelligence, functioned as a Business Analyst at EarlyNinja, and provided freelance support as an Assistant at Pro Web Consulting.

Highlights

Aug 28 · linkedin

Despite its relevance, the concept of data leakage is little mentioned compared to overfitting. This makes it even more dangerous for the emerging data scientist.

Aug 27 · linkedin

Why automate internet browsing with natural language? Well the answer is simple: to allow non-technical people to create macros and automations through a common interface.

Many of these tasks are repetitive, time-consuming and require little to no cognitive effort.

By automating these tasks, #LaVague aims to free up time for more meaningful activities, allowing users to focus on what really matters to them.

By providing an engine that transforms natural language queries into Selenium code, LaVague is designed to allow users or other #AIs to easily automate expressed web workflows and execute them in a browser.

Here's what LaVague can do for you:

  • Understands natural language instructions to perform browser interactions

  • Integrates with Selenium, a popular browser emulator used by programmers to test and automate development, to automate web browsers.

  • Builds on open source projects like Transformers and #LlamaIndex and leverages open source models, both locally and remotely, to ensure the agent is transparent and aligned with user interests

  • Uses local embedding models to run #RAG and extract the most relevant HTML parts to feed the LLM that answers the query

Behind the scenes, LaVague leverages few-shot learning and chain of thought thinking to get the most relevant Selenium code to perform the action without having to tune the LLM for code generation.

One of the key uses is to automate personal user tasks and require them to log in, such as automating the process of paying bills, filling out forms, or extracting data from specific websites.

You can check LaVague's repo here: https://lnkd.in/dPHqEgkW

Andrea D'Agostino
Add to my network

Location

Greater Naples Metropolitan Area