Luiz Felipe Mendes

Leading Recommendation, Ads, and Search Data Science teams (>20 data scientists).

Built AI-driven platforms powering personalization, search ranking, and discovery for millions of users.Collaborated with engineering, product, and business teams to drive data-driven decision-making.

Key Projects & Achievements:

• FoodGraph – Developed a graph-based deep learning model using user order history to

generate embeddings, enhancing recommendation discovery and engagement.

• Recommendation Platform – Designed a scalable hybrid recommendation system,

integrating:

• Candidate generation pipelines for surfacing relevant options.
• Offline models for ranking and personalization.
• Re-ranking techniques to optimize engagement and conversion rates.
• Semantic Search – Built domain-specific embeddings for food and grocery delivery, improving query understanding and item representation, enhancing search relevance.

• Co-founded Hekima, initially focused on social media analytics, later expanding into

enterprise AI & Big Data solutions.

• Led a Data Science team developing machine learning & data analytics projects for

major Brazilian companies.

• Transitioned to iFood after Hekima was acqui-hired in 2020, leading AI-driven

personalization efforts.

Key Projects & Achievements:

• Led AI & Data Science Teams – Managed a multidisciplinary team of data scientists, engineers, and analysts, driving end-to-end AI solutions.
• Enterprise AI Solutions – Delivered machine learning & data analysis projects for clients like AMBEV, Raízen, Fiat, and JCPM Group, focusing on NLP, optimization, e-commerce, and data infrastructure.
• Zahpee Monitor – Co-developed an AI-powered social media analytics platform, analyzing millions of social media posts in real-time, enabling businesses to track sentiment and trends.
• AI Education & Training – Created and taught Data Science Course on Udemy, educating professionals on machine learning, AI applications, and data-driven decision-

making.

Specialization in NLP focused on Text Classification and Meta Classifiers