MLOps.community A Playground for AI/ML Engineers
18 snips
Jan 23, 2026 Paulo Vasconcellos, Principal Data Scientist for Generative AI Products at Hotmart and co-founder of Data Hackers, builds AI tools for creators and learners. He discusses using LLMs alongside classic NLP for speed and cost. He explores multilingual model choices, agent-as-a-product creator tools, Hotmart Tutor as a 24/7 teacher, and engineering tradeoffs for scalable ML and safety guardrails.
AI Snips
Chapters
Transcript
Episode notes
Hotmart As A Data Science Playground
- Hotmart is a playground for data scientists solving fraud, recommendations, and forecasting since 2015.
- Paulo described early production ML work including fraud detection and recommender systems that launched years ago.
LLMs Don't Obliterate Classic NLP
- LLMs are transformative but don't replace all classical NLP models for performance, cost, and latency reasons.
- Paulo noted entity extraction and high-volume ticket classification remain cheaper and faster with classical models.
Weigh Tradeoffs Before Replacing Models
- Evaluate tradeoffs before replacing classical models with LLMs, weighing training cost, drift, and infrastructure.
- Maintain pipelines for continuous training and monitoring if you choose classical approaches.