> Markdown version of [/videos/1305-building-your-own-classification-model-with-javascript-coffee-with-developers-carly-richmond?t=1861](https://www.wearedevelopers.com/videos/1305-building-your-own-classification-model-with-javascript-coffee-with-developers-carly-richmond?t=1861). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Building Your Own Classification Model with JavaScript - Coffee with Developers - Carly Richmond

Carly Richmond warns that off-the-shelf AI plugins stifle developer curiosity. Discover how she built a TensorFlow.js classification model to prove the lasting value of hands-on tinkering.

- **Speakers:** Carly Richmnd
- **Event:** Coffee With Developers
- **Published:** March 5, 2025
- **Duration:** 36:29
- **URL:** https://www.wearedevelopers.com/videos/1305-building-your-own-classification-model-with-javascript-coffee-with-developers-carly-richmond

## Summary

Carly Richmond, Principal Developer Advocate at Elastic, shares her journey of diving into machine learning as a web developer by building a JavaScript-based binary classification model using TensorFlow.js. Inspired by the Netflix show *Is It Cake?*, her initial attempt to build a custom model from scratch struggled with false positives due to limited dataset sizing and dimensionality. However, by pivoting to transfer classification with the pre-trained MobileNet model, she significantly improved the system's success rate, proving the immense value of experimenting iteratively and embracing failure to deepen technical understanding.

Beyond the technical implementation, the discussion highlights how modern fast-paced software environments have shifted developers toward consuming off-the-shelf AI plugins rather than building their own. Because intense sprint deadlines often stifle natural curiosity, it is critical for organizations to prioritize unstructured tinkering time—such as intentional gap days, hackathons, or dedicated experimental sprints like "Space Time Weeks." Without this protected free time backed by engineering leadership, developers risk losing the deep, hands-on competency required to effectively assess and troubleshoot complex architectural features.

The conversation also probes the limitations, ethics, and authenticity of contemporary AI. Richmond underscores the necessity of detailed "model cards" on platforms like Hugging Face to transparently expose training biases and dataset constraints. Furthermore, as the hype around automated content generation escalates, she emphasizes the enduring value of human authenticity. AI-generated tutorials inherently lack the nuanced developer experience, anecdotal humor, and genuine problem-solving struggle, cementing the software engineer's future role as both an empathetic storyteller and a rigorous fact-checker of AI output.

**Keywords:** javascript classification models, tensorflow.js image detection, binary classifier training, transfer classification methodologies, mobilenet model implementation, machine learning dataset limitations, hugging face model cards, ai build versus buy debate, developer hackathon culture, engineering gap days, ai false positives debugging, generative ai authenticity, developer advocacy strategies, ai training bias transparency

## Chapters

1. **Experimenting with JavaScript machine learning for cake detection** (00:20) — Building an image classification game using TensorFlow to identify objects disguised as cakes.
1. **Comparing custom machine learning models with transfer classification** (02:57) — How using pre-trained models like MobileNet improves accuracy over building custom binary classifiers from scratch.
1. **Impact of prebuilt artificial intelligence on developer curiosity** (07:25) — Relying on pre-built AI models out of convenience reduces learning opportunities for software engineers.
1. **Carving out dedicated time for engineering side projects** (11:21) — How engineering leaders can intentionally schedule gap days to retain coding skills and technical depth.
1. **Maintaining the value of hackathons and unstructured exploration** (15:15) — Management buy-in and clear outcomes are necessary to prevent developers from abandoning internal hack days.
1. **Evaluating trust and transparency in machine learning models** (19:52) — Why standardized model cards are crucial for understanding the dataset limitations and biases of pre-trained models.
1. **How dominant language models change developer search behavior** (22:59) — The affordability barrier of custom datasets pushes engineers toward off-the-shelf artificial intelligence tools without validating responses.
1. **Preserving authenticity in technical content and documentation writing** (26:39) — Why the surge of automated content generation shifts the engineering role toward fact-checking and reviewing code.
1. **Balancing artificial intelligence automation with human connection and ethics** (31:01) — The importance of establishing ethical usage policies to combat deep fakes and preserve human storytelling.

## Related Moments

- [Differentiating developer skills in the era of artificial intelligence](https://www.wearedevelopers.com/videos/1753-wearedevelopers-live-spicy-vanilla-web-css-magic-more) (from "WeAreDevelopers LIVE – Spicy Vanilla Web, CSS Magic & More")
- [Addressing developer burnout and the impact of artificial intelligence](https://www.wearedevelopers.com/videos/1504-cracking-the-code-to-tech-team-satisfaction) (from "Cracking the Code to Tech Team Satisfaction")
- [Addressing psychological safety and ethical risks of AI adoption](https://www.wearedevelopers.com/videos/1950-the-scrum-master-as-an-orchestrator-guiding-human-ai-collaboration-in-modern-teams) (from "The Scrum Master as an Orchestrator: Guiding Human–AI Collaboration in Modern Teams")
- [Balancing AI enthusiasm with cynical engineering tool practices](https://www.wearedevelopers.com/videos/1858-a-stack-overflow-for-agents-peter-wilson) (from "A Stack Overflow for Agents? - Peter Wilson")
- [Addressing developer burnout and the perceived value of AI](https://www.wearedevelopers.com/videos/100093-the-integrated-ai-experience-a-new-paradigm-in-an-agentic-world) (from "The integrated AI experience: A New Paradigm in an Agentic World")
- [Preserving the joy and curiosity of human coding](https://www.wearedevelopers.com/videos/1764-wearedevelopers-live-web-scraping-agents-actors-and-more) (from "WeAreDevelopers LIVE – Web Scraping, Agents, Actors and more")

## Related Articles

- [Exploring AI: Opportunities and Risks for Developers](https://www.wearedevelopers.com/magazine/522-exploring-ai-opportunities-and-risks-for-developers)
- [WWC24 Talk - Scott Hanselman - AI: Superhero or Supervillain?](https://www.wearedevelopers.com/magazine/469-wwc24-talk-scott-hanselman-ai-superhero-or-supervillain)
- [What is Software Engineering in the Age of AI?](https://www.wearedevelopers.com/magazine/640-what-is-software-engineering-in-the-age-of-ai)
- [Dev Digest 137 - AI'm not sure about this](https://www.wearedevelopers.com/magazine/485-dev-digest-137-ai-m-not-sure-about-this)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Machine Learning Engineer](https://www.wearedevelopers.com/jobs/ext/588393-machine-learning-engineer) at **Twilio**
- [Staff Developer Advocate, GitHub Security Lab](https://www.wearedevelopers.com/jobs/ext/1921051-staff-developer-advocate-github-security-lab) at **GitHub**
- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [Machine Learning Engineer](https://www.wearedevelopers.com/jobs/ext/1355348-machine-learning-engineer) at **TWILIO**