Fireside chat #1: How to Produce Sustainable Business Value with Machine Learning

Outerbounds · Beginner ·📋 Product Management ·4y ago

Key Takeaways

The video discusses product management for machine learning-powered products, focusing on generating sustainable business value with ML, with expert insights from Hilary Parker and Hugo Bowne-Anderson, covering topics such as when to adopt ML, common failure modes, and full-stack machine learning.

Original Description

Hilary Parker, a data scientist and independent consultant who has worked at Stitch Fix, Etsy, and as a data science product manager for the 2020 Biden for President Campaign, joins Hugo Bowne-Anderson and Outerbounds to talk about product management (PM) for products powered by machine learning (ML) and how to produce sustainable business value with ML. The modern capabilities of data science and machine learning are wonderful but, as an industry, we’re still figuring out how to generate sustainable business value with them, for the majority of businesses and industries. In this fireside chat, Hilary and Hugo will discuss what ML is useful for (and what it isn’t!), when companies should adopt ML, and what failure modes of ML-powered products we should all keep in mind. 00:00:00 Prelude 00:04:45 We begin to chat! 00:06:15 Hilary's background in data science and product management 00:14:15 Data at Stitchfix and Netflix 00:22:10 When should companies adopt ML? 00:30:00 ML Failure modes 00:40:10 Product management for ML products 00:50:20 On full-stack machine learning 00:58:45 What do PMs need to know about data science and vice versa? 01:02:20 Is the future of all software machine learning? Find out more about how we think about MLOps, OSS, and human-centric data science tools here: https://outerbounds.com/
Watch on YouTube ↗ (saves to browser)
Sign in to unlock AI tutor explanation · ⚡30

Playlist

Playlist UU5h8Ji6Lm1RyAZopnCpDq7Q · Outerbounds · 3 of 60

1 Metaflow GUI for monitoring machine learning workflows
Metaflow GUI for monitoring machine learning workflows
Outerbounds
2 Metaflow Cards [no sound]
Metaflow Cards [no sound]
Outerbounds
Fireside chat #1: How to Produce Sustainable Business Value with Machine Learning
Fireside chat #1: How to Produce Sustainable Business Value with Machine Learning
Outerbounds
4 Fireside chat #2: MadeWithML.com -- Teaching Practical Machine Learning
Fireside chat #2: MadeWithML.com -- Teaching Practical Machine Learning
Outerbounds
5 Metaflow on Kubernetes and Argo Workflows [no sound]
Metaflow on Kubernetes and Argo Workflows [no sound]
Outerbounds
6 Fireside chat #3: Reasonable Scale Machine Learning -- You're not Google and it's totally OK
Fireside chat #3: Reasonable Scale Machine Learning -- You're not Google and it's totally OK
Outerbounds
7 Metaflow Tags: Programmatic Tagging
Metaflow Tags: Programmatic Tagging
Outerbounds
8 Metaflow Tags: Basic Tagging
Metaflow Tags: Basic Tagging
Outerbounds
9 Metaflow Tags: Tags in CI/CD
Metaflow Tags: Tags in CI/CD
Outerbounds
10 Metaflow Tags: Tags and Namespaces
Metaflow Tags: Tags and Namespaces
Outerbounds
11 Metaflow Tags: Tags and Continuous Training
Metaflow Tags: Tags and Continuous Training
Outerbounds
12 Fireside chat #4: Machine Learning and User Experience -- Building ML Products for People
Fireside chat #4: Machine Learning and User Experience -- Building ML Products for People
Outerbounds
13 Fireside Chat #5: Machine Learning + Infrastructure for Humans
Fireside Chat #5: Machine Learning + Infrastructure for Humans
Outerbounds
14 Metaflow Sandbox Demo: Free Data Science Infrastructure In the Browser
Metaflow Sandbox Demo: Free Data Science Infrastructure In the Browser
Outerbounds
15 Metaflow on Azure
Metaflow on Azure
Outerbounds
16 Fireside Chat #6: Operationalizing ML -- Patterns and Pain Points from MLOps Practitioners
Fireside Chat #6: Operationalizing ML -- Patterns and Pain Points from MLOps Practitioners
Outerbounds
17 ML engineering vs traditional software engineering: similarities and differences
ML engineering vs traditional software engineering: similarities and differences
Outerbounds
18 Why data scientists love and hate notebooks: velocity and validation
Why data scientists love and hate notebooks: velocity and validation
Outerbounds
19 What even is a 10x ML engineer?
What even is a 10x ML engineer?
Outerbounds
20 The 4 main tasks in the production ML lifecycle
The 4 main tasks in the production ML lifecycle
Outerbounds
21 Is the premise of data-centric AI flawed?
Is the premise of data-centric AI flawed?
Outerbounds
22 The 3 factors that Determine the success of ML projects
The 3 factors that Determine the success of ML projects
Outerbounds
23 Fireside Chat #7: How to Build an Enterprise Machine Learning Platform from Scratch
Fireside Chat #7: How to Build an Enterprise Machine Learning Platform from Scratch
Outerbounds
24 Run Metaflow on any cloud: Google Cloud, Azure, or AWS [no sound]
Run Metaflow on any cloud: Google Cloud, Azure, or AWS [no sound]
Outerbounds
25 Metaflow on GCP
Metaflow on GCP
Outerbounds
26 Fireside Chat #8: Navigating the Full Stack of Machine Learning
Fireside Chat #8: Navigating the Full Stack of Machine Learning
Outerbounds
27 How to Build a Full-Stack Recommender System
How to Build a Full-Stack Recommender System
Outerbounds
28 Modernize your Airflow deployments with Metaflow - zero-cost migration [no sound]
Modernize your Airflow deployments with Metaflow - zero-cost migration [no sound]
Outerbounds
29 Easy Airflow DAGs for ML and data science with Metaflow [no sound]
Easy Airflow DAGs for ML and data science with Metaflow [no sound]
Outerbounds
30 Fireside chat #9:  Language Processing: From Prototype to Production
Fireside chat #9: Language Processing: From Prototype to Production
Outerbounds
31 How to build end-to-end recommender systems at reasonable scale
How to build end-to-end recommender systems at reasonable scale
Outerbounds
32 Full-Stack Machine Learning with Metaflow on CoRise
Full-Stack Machine Learning with Metaflow on CoRise
Outerbounds
33 Natural Language Processing meets MLOps
Natural Language Processing meets MLOps
Outerbounds
34 Fireside Chat #10: Large Language Models: Beyond Proofs of Concept
Fireside Chat #10: Large Language Models: Beyond Proofs of Concept
Outerbounds
35 What even are Large Language Models?
What even are Large Language Models?
Outerbounds
36 How to get started with LLMs today
How to get started with LLMs today
Outerbounds
37 LLMs in production
LLMs in production
Outerbounds
38 Accessing secrets securely in Metaflow [no audio]
Accessing secrets securely in Metaflow [no audio]
Outerbounds
39 Fireside Chat #11: The Open-Source Modern Data Stack
Fireside Chat #11: The Open-Source Modern Data Stack
Outerbounds
40 Fireside chat #12: Kubernetes for Data Scientists
Fireside chat #12: Kubernetes for Data Scientists
Outerbounds
41 Behind the Screen: How Amazon Prime Video ships RecSys models 4x faster
Behind the Screen: How Amazon Prime Video ships RecSys models 4x faster
Outerbounds
42 Fireside chat #13: Supply Chain Security in Machine Learning
Fireside chat #13: Supply Chain Security in Machine Learning
Outerbounds
43 Quick Delivery, Quicker ML: DeliveryHero's Metaflow Story
Quick Delivery, Quicker ML: DeliveryHero's Metaflow Story
Outerbounds
44 Crafting General Intelligence: LLM Fine-tuning with Metaflow at Adept.ai
Crafting General Intelligence: LLM Fine-tuning with Metaflow at Adept.ai
Outerbounds
45 Fuelling Decisions: How DTN Powers Gas Pricing and Data Science Collaboration
Fuelling Decisions: How DTN Powers Gas Pricing and Data Science Collaboration
Outerbounds
46 From Kitchen to Doorstep: Optimizing Data Science Velocity at Deliveroo
From Kitchen to Doorstep: Optimizing Data Science Velocity at Deliveroo
Outerbounds
47 Building a GenAI Ready ML Platform with Metaflow at Autodesk
Building a GenAI Ready ML Platform with Metaflow at Autodesk
Outerbounds
48 Media Transcoding for 10 Million users and beyond with Metaflow at Epignosis
Media Transcoding for 10 Million users and beyond with Metaflow at Epignosis
Outerbounds
49 Telematics with Metaflow: How Nirvana Insurance built a large-scale Risk Estimation platform
Telematics with Metaflow: How Nirvana Insurance built a large-scale Risk Estimation platform
Outerbounds
50 Fireside chat #14: Generative AI and Machine Learning for Film, TV, and Gaming
Fireside chat #14: Generative AI and Machine Learning for Film, TV, and Gaming
Outerbounds
51 The Past, Present, and Future of Generative AI
The Past, Present, and Future of Generative AI
Outerbounds
52 Building Production Systems with Generative AI, Machine Learning, and Data
Building Production Systems with Generative AI, Machine Learning, and Data
Outerbounds
53 A Custom Fine-Tuned LLM in Action (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 5)
A Custom Fine-Tuned LLM in Action (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 5)
Outerbounds
54 Building Live Production Systems with RAG (LLMs & RAG: An Interactive Guided Tour Part 4)
Building Live Production Systems with RAG (LLMs & RAG: An Interactive Guided Tour Part 4)
Outerbounds
55 Better Relevancy with RAG (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 3)
Better Relevancy with RAG (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 3)
Outerbounds
56 Working with OSS LLMs (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 2)
Working with OSS LLMs (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 2)
Outerbounds
57 Hitting OpenAI and Other Vendor APIs (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 1)
Hitting OpenAI and Other Vendor APIs (LLMs, RAG, and Fine-Tuning: An Interactive Guided Tour Part 1)
Outerbounds
58 Production Systems with Generative AI (LLMs, RAG, & Fine-Tuning: An Interactive Guided Tour Part 0)
Production Systems with Generative AI (LLMs, RAG, & Fine-Tuning: An Interactive Guided Tour Part 0)
Outerbounds
59 LLMs in Practice: A Guide to Recent Trends and Techniques
LLMs in Practice: A Guide to Recent Trends and Techniques
Outerbounds
60 Metaflow for distributed high-performance computing and large-scale AI training
Metaflow for distributed high-performance computing and large-scale AI training
Outerbounds

This video provides insights into product management for machine learning-powered products, covering topics such as ML adoption, failure modes, and full-stack machine learning, with expert advice from Hilary Parker and Hugo Bowne-Anderson. Viewers will learn how to generate sustainable business value with ML and develop effective product strategies for ML-powered products. The discussion highlights the importance of human-centric data science and MLOps in producing successful ML-powered products

Key Takeaways
  1. Determine when to adopt ML for your business
  2. Identify potential ML failure modes and develop strategies to avoid them
  3. Develop a product strategy for your ML-powered product
  4. Focus on full-stack machine learning to ensure successful product development
  5. Collaborate with data scientists and product managers to ensure human-centric data science and MLOps
  6. Stay up-to-date with the latest developments in ML and data science
💡 The key to generating sustainable business value with ML is to focus on full-stack machine learning, avoid common failure modes, and develop effective product strategies that prioritize human-centric data science and MLOps.

Related AI Lessons

Your PM Can Now Ship Without a Designer. Here's When That's Stupid.
Learn when a product manager should not ship without a designer's input, despite AI advancements
Hackernoon
Which Agency Helped the Most Successful AI Startups With Product Design?
Learn how product design contributed to the success of top AI startups and which agency helped them achieve it
Medium · Startup
117 People Waited 21 Years for One Feature. It Finally Arrived.
Learn how perseverance and determination can lead to the implementation of a long-awaited feature despite initial rejections from experts
Medium · Programming
The Blurring Lines of Product Design: What Happens When Your Role Expands Beyond the Screen
Learn how product design roles are expanding beyond screen design to encompass business and stakeholder considerations
Medium · UX Design

Chapters (10)

Prelude
4:45 We begin to chat!
6:15 Hilary's background in data science and product management
14:15 Data at Stitchfix and Netflix
22:10 When should companies adopt ML?
30:00 ML Failure modes
40:10 Product management for ML products
50:20 On full-stack machine learning
58:45 What do PMs need to know about data science and vice versa?
1:02:20 Is the future of all software machine learning?
Up next
One Wing Won't Fly | Gurmukh Singh Bawa | TEDxDWPS Ludhiana Youth
TEDx Talks
Watch →