Adventures in Machine Learning

Jesse Langford spent the first half of his career as a golf instructor before pivoting to software engineering.  Today on the show, Ben interviews Jesse to learn why and how he made this pivot, plus relevant career advice for all developers.  Specific topics include taking ownership of your work, being comfortable making mistakes, and how to stretch yourself every day.

In this episode…

1. Taking ownership
2. Being comfortable taking risks and making mistakes
3. From back end to front end development
4. Self taught and self paced
5. Dunning-Kruger effect
6. Continually growing and learning

Sponsors

• Top End Devs
• Coaching | Top End Devs

Links

• The Most Important Thing I Did to Become a Senior Developer
• Jesse Langford
• LinkedIn: Jesse Langford

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

When developing ML models, defining and selecting the model architecture will be fundamental to ensure the best possible outcomes.  Parameters that define the model architecture are referred to as hyperparameters and the process of searching for the ideal model architecture is referred to as hyperparameter tuning.  Today on the show, Ben and Michael discuss hyperparameter tuning and how to implement this into your ML modeling.

In this episode…

1. Why do we tune?
2. Optimizing the models
3. Hyperparameter tuning
4. Steps for tuning
5. Data splits
6. Linear based models
7. How do you know when you know enough?
8. Basic rules of thumb
9. Buffer in time for spikes
10. Grid searching and automation

Sponsors

• Top End Devs
• Coaching | Top End Devs

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

Enjoy this engaging AMA conversation with Michael Berk asking Ben Wilson various questions related to industry, strategy, and approaches in data science and ML engineering.
In this episode…

1. Why should people trust you?
2. What will you lose by hearing about other people’s failures vs. personally failing to learn?
3. How do you view the current industry?
4. Do you think data scientists and ML engineers are overpaid?
5. What are 3 things important to ROI?
6. How do you set the tone for culture?

Sponsors

• Top End Devs
• Coaching | Top End Devs

Links

• Machine Learning Engineering in Action

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

Ben and Michael interview Maciej Balawejder, a mechanical engineering student passionate about AI, ML, and robotics.  As an active contributor on Medium.com, Maciej has already made significant contributions to the AI and ML communities.  On the show, they discuss Maciej’s recent article about optimizers in Machine Learning, plus their personal philosophies and approaches to deep learning.

Sponsors

• Top End Devs
• Coaching | Top End Devs

Links

• Maciej Balawejder - Medium
• Optimizers in Machine Learning

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

After ensuring your data has surpassed the hyper parameter tuning phase, what is the next step in your EDA protocol? Today on the show, Ben and Michael continue the discussion on EDA methodology within Machine Learning and discuss linear regression with OLS, decision trees, and common visualization tools for data scientists.

In this episode...

1. Linear regression with OLS
2. Accuracy metrics
3. Decision trees
4. Shapley values
5. Common visualization tools

Sponsors

• Top End Devs
• Coaching | Top End Devs

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

EDA is primarily used in machine learning to see what data can reveal beyond the formal modeling or hypothesis testing task and provides a better understanding of data set variables and the relationships between them. It can also help determine if the statistical techniques you are considering for data analysis are appropriate. Today on the show, Ben and Michael discuss how to use EDA in machine learning models.

In this episode...

1. What is EDA?
2. Tips and Tricks and steps for EDA
3. How to approach downsampling
4. Understanding feature sets relative to your labels
5. Optimizing models
6. Motivating yourself to get into the data
7. Tools for EDA
8. A few scenarios for discussion
9. What is the most detrimental EDA mistake for ML

Sponsors

• Top End Devs
• Coaching | Top End Devs

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

MLlib is Apache Spark's scalable machine learning library. Today, Ben and Michael discuss the ease of use, performance, algorithms, and utilities included in this library and how to execute the best ML workflow with MLlib.

In this episode...

1. Why stick with Spark libraries vs. a single node operation?
2. What algorithms are not in Spark Lib?
3. What is the min. package set to use for supervised learning?
4. Modeling and validation
5. Down-sampling your data
6. MLlib vs. scikit-learn
7. Resources

Sponsors

• Top End Devs
• Coaching | Top End Devs

Links

• MLlib | Apache Spark
• What is PySpark? | Domino Data Science Dictionary
• UCI Machine Learning Repository

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

Apache Spark is a lightning-fast unified analytics engine for large-scale data processing and machine learning. In this episode, Ben and Michael unpack Spark by ping-ponging questions and answers, supplemented by various examples applicable to machine learning workflows. In this Episode…

1. How does Spark work?
2. What makes Apache Spark effective?
3. Dot repartition in Spark
4. Parallel processing systems
5. What is an aggregation in Spark sequel?
6. Analytics with Spark
7. What is MPP?
8. Testing for production
9. Spark algorithms

Sponsors

• Top End Devs
• Coaching | Top End Devs

Sponsored By:

• Coaching | Top End Devs: Do you want to level up your career? or go freelance? or start a podcast or youtube channel? Let Charles Max Wood Help You Achieve Your Dreams
• Top End Devs: Learn to Become a Top 5% Developer. Join our community of ambitious and engaged programmers to learn how.

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

Ben and Michael walk through two different cases studies relative to production ML infrastructure and recommendation engines. The first is about a free on-line tutoring service for underserved communities called “Learn to Be”, and the second centers around the online course provider “Coursera”. Ben and Michael set up the case studies with fundamental problem statements, followed by their various approaches to executing the objectives to achieve the desired process outcomes. Sponsors

• Top End Devs
• Coaching | Top End Devs

Sponsored By:

• Coaching | Top End Devs: Do you want to level up your career? or go freelance? or start a podcast or youtube channel? Let Charles Max Wood Help You Achieve Your Dreams
• Top End Devs: Learn to Become a Top 5% Developer. Join our community of ambitious and engaged programmers to learn how.

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.

Ben interviews Michael Griffiths, Director of Data Science at ASAPP, a company leveraging AI and ML to augment and automate human work, improve operational efficiencies and customer experiences, and ultimately empower people to be their best. Michael shares specific examples of how this can be done for human agent productivity within contact centers. They also discuss fully human controlled vs automated systems, delivering value with AI and ML, and the future of AI driven technology. In this Episode…

1. How do you deliver value with AI and ML?
2. Fully human controlled vs. fully automated systems
3. ML software engineering vs traditional software
4. Using static training data models and data validation
5. Communicating process improvements and failures
6. What is the future of ASAPP and AI driven technology?

Sponsors

• Top End Devs
• Coaching | Top End Devs

Links

• How a Level System can Help Forecast AI Costs - KDnuggets
• AI Research - ASAPP
• ASAPP

Special Guest: Michael Griffiths.Sponsored By:

• Coaching | Top End Devs: Do you want to level up your career? or go freelance? or start a podcast or youtube channel? Let Charles Max Wood Help You Achieve Your Dreams
• Top End Devs: Learn to Become a Top 5% Developer. Join our community of ambitious and engaged programmers to learn how.

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

Become a supporter of this podcast: https://www.spreaker.com/podcast/adventures-in-machine-learning--6102041/support.