Mattia Gaggi

Machine Learning Engineer/Founder - ex Tractable @Tune AI

🔥 Startup Founder✔️ IT🧑‍🔬 EngineerFounderComputer engineerML EngineerData Scientist

I am an experienced Applied Researcher with a customer-centric approach that has delivered multiple systems from concept to maintenance mode with off-the-shelf and novel Computer Vision and NLP components. I partner across Product, Sales, Research, and Engineering for requirements scoping, system design, data labeling, and deploying models in production. These systems scaled to nationwide deployment for large enterprise customers in highly regulated and safety-critical environments. Previously worked at Tractable (Europe's first AI unicorn) and Optellum (Spin-off of Oxford University's VGG AI Lab). I build end-to-end custom solutions and I am used to working on any Machine Learning platform.

Feedback from Mattia’s ex-colleagues

Let me see

Keep in touch with meI'm using Intch to connect with new people. Use this link to open chat with me via Intch app

Work Background

Entrepreneur FirstFounder in Residence (LD20)

Mar. 2023London, England, United Kingdom · On-siteTesting co-founding partnerships and ideas with the world’s most ambitious people.Selected from 1000+ applicants to join Entrepreneur First’s platform and work with a community of entrepreneurs and leading advisors to find a co-founder, and build a company from scratch. Developed and launched a Generative AI platform that converted text descriptions into printable 3D models. The platform was launched on Discord and obtained 1000 MAU. Used state-of-the-art Generative AI models such as Stable Diffusion and NeRF to generate 3D models from text descriptions. The system was built using Discord as a UI and SQS to send messages to an auto-scaling EC2 GPU cluster. Gained valuable business and product experience in launching a Generative AI platform. The product has been discontinued as there was not a viable business model for our customer base.

Tune AIF0under

Oct. 2017London, England, United KingdomMy own consulting company through which I have done contracting ( odd characters in the title are to avoid marketing emails)

MOONSIFTMachine Learning Lead

Feb. 2024 - Jun. 2024London Area, United Kingdom · RemoteMoonsift is a platform for ecommerce: users add any fashion product to their wish list from any website and can search similar items. Working directly with the CTO, I am shaping the roadmap for implementing recommendation models and improve the main search product as well as defining best practices for the company. I developed an algorithm for ranking users’ favourite stores and productionized it with the purpose of ranking items. I planned and implemented the architecture end-to-end and training models such as Matrix Factorization, Autoencoders, Multimodal models (CLIP) on data from users.

Autoenhance.aiHead of Artificial Intelligence

May. 2023 - Nov. 2023London, England, United Kingdom · On-siteAutoenhance automates photo editing done by real-estate agents and their subcontractors, saving time & money. Working directly with the CEO, Product Managers, and Engineers, I defined the research roadmap and led ML development and deployment. I developed novel computer-vision models and techniques to enable the product to autonomously perform Multi-exposure Image Fusion (MEF), Color Correction, and Perspective Correction on images.Deployed the model on AWS Sagemaker. To achieve high-reliability, I led labeling and training on in-house datasets. These improvements unlock partnerships with professional editing companies which will increase ARR by 30%.

TractableMachine Learning Research Engineer

Dec. 2021 - Sep. 2022London, England, United KingdomTractable is a Computer Vision unicorn for car accidents and property damage appraisals. Tractable is an ML unicorn that automates claims processing for top-tier global insurers. Our team enabled customers to process paper documents with Tractable using Computer Vision and NLP techniques. I collaborated with customer-facing Product, Engineering, and Sales teams to lead delivery of new ML features and systems from concepts to maintenance mode, including novel methods for document classification, text-recognition, and Named Entity Recognition (NER) using LLMs (RoBERTa), segmentation, and other visual models. These systems scaled to nationwide deployment at multiple top-tier global insurers, creating expansion opportunities in North America, enabling retention of 45% ARR, and saving internal costs >$3mn/year.

OptellumMachine Learning Researcher

Dec. 2019 - Nov. 2021Oxford, England, United KingdomOptellum is a spin-off from the world-class AI lab, Oxford University Visual Geometry Group (VGG). Contributed to the first FDA & CE approved ML models for early lung cancer diagnosis, enabling earlier diagnosis and improved patient outcomes after being deployed to several hospitals in the US. My responsibilities include improving the current ML products, researching and implementing novel ML classification and segmentation algorithms for future products.

Intelligent DatalyticsData Scientist - Independent Contractor

Oct. 2017 - Sep. 2018Rome Area, ItalyDeveloped a text web-scraping software(run on a local machine) to gather patenting data and saved 100+ hours of manual work for an industry leading Pharma company. Developed a PoC churn prediction Machine Learning model, which demonstrated a 3% reduction in churn for a telecommunications company with ~18m customers. Successfully deployed the model in 3 offices as a proof of concept and presented the work to senior executives to spearhead its adoption across the entire organization.