Creating Tools that Spark Joy with Ines Montani

2022-01-27 · Source: Explosion · Developer tools and consulting for AI, Machine Learning and NLP - Explosion.ai · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, extended

Summary

Ines Montani, co-founder and CEO of Explosion, the Berlin-based company behind the spaCy NLP library and Prodigy data annotation tool, discusses their philosophy for creating user-centric developer tools. Explosion prioritizes a holistic user experience, designing APIs for extensibility and focusing on high ceilings for complex tasks, not just ease of initial use. Montani emphasizes interpreting user feedback to address underlying problems rather than just implementing requested features, advocating for a "let them write code" approach to support niche use cases. The discussion also covers the design process at Explosion, which involves extensive planning documents for their remote-first team. Regarding NLP in production, Montani defines it as a system providing value, noting the landscape's growing maturity but highlighting challenges in continuous model iteration, data management, and reproducibility. She details spaCy 3's config file and project system, designed to ensure reproducible workflows and bridge the gap between prototyping and production.

Key takeaway

For ML Engineers building production-grade NLP systems, prioritize a unified workflow that bridges prototyping and deployment. Your tools should offer extensible APIs, like spaCy's "let them write code" philosophy, enabling custom solutions without waiting for features. Embrace continuous iteration on models and data, recognizing that NLP systems are never truly "done." Leverage features like spaCy 3's config file and project system to ensure reproducibility and streamline end-to-end processes, avoiding the pitfalls of disparate development environments.

Key insights

User-centric developer tools thrive on deep problem understanding, extensibility, and integrated production workflows, ensuring joy and productivity.

Principles

• Design for extensibility via programmable APIs.
• Interpret user feedback to solve core problems.
• Ensure tools offer easy start, high ceiling.

Method

Explosion's design process involves collaborative prototyping and extensive planning documents for their remote team, emphasizing attention to detail and consistency. spaCy 3's config file and project system orchestrate reproducible end-to-end NLP workflows, from data preprocessing to deployment.

In practice

• Utilize spaCy 3's config for reproducible settings.
• Adopt spaCy's project system for workflow orchestration.
• Establish one workflow for prototyping and production.

Topics

• Natural Language Processing
• spaCy
• Prodigy
• MLOps
• Reproducibility
• API Design

Best for: Machine Learning Engineer, NLP Engineer, MLOps Engineer

Related on AIssential

• Evolution of spaCy — Industry-grade NLP requires opinionated, fast, and extensible tools that prioritize developer engagement with data.
• The triangulation of ethical leader signals using qualitative, experimental, and data science methods — A complete spaCy-based NLP project workflow integrates data, annotation, training, and deployment.
• Getting My Hands Dirty with Python and NLP — Building a practical NLP project is an effective way to learn Python libraries and data structures.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Explosion · Developer tools and consulting for AI, Machine Learning and NLP - Explosion.ai.