Module 2 - LO3 Human Resources Roles and Functions.pptx
Staff Machine Learning Engineer Performance Based Job Description
1. Staff Machine Learning Engineer
Ramping Up
● In your first 2 weeks, you’ll learn about the Woebot content architecture and how ML and NLP are
used to guide conversations with our users.
● In your first 3 weeks, you’ll list improvements that could be made to our existing set of classifiers.
Own Our Machine Learning Models, Systems & Processes
● During your first 45 days you will develop infrastructure for the full cycle of our machine learning
efforts, this includes, model training, feature extraction, deploying produced models, data processing,
and rigorously A/B testing.
○ To accomplish this you'll collaborate with engineers to integrate algorithms efficiently with
backend production service and help define our data team's processes and tooling.
● You will also build machine learning models that enable Woebot to more naturally understand users’
natural language input, and generate appropriate responses. Initially you will prioritize the following:
○ Improve Woebot’s Sentiment Analysis and work with our Product team to define solutions and
integrate them into a 1-year roadmap.
○ Help scale our services using GPUs and modern distributed processing tools in the cloud
(AWS).
○ Dig into data with ad hoc analysis as necessary for technical, clinical, and user needs.
Help Woebot Have More Natural Flowing Conversation
● Within your first 60 days you will be responsible for gathering data and building data labeling
systems so that Woebot continually learns from ongoing conversations to better recognize the
sentiment and intent behind users’ natural language inputs.
● You will have unique datasets such as: 1M+ user conversations, support tickets, and other
natural/unstructured data sources
Turn Millions of Data Points Into Valuable Insights
● By day 90 you will create user profiles and build models that define how Woebot interacts with each
user. We consider personalization to be key for developing a relationship over time, and delivering
precision intervention - that is, methods that are tailored to each user.
● You will improve our machine learning models that enable Woebot to more naturally converse with
and understand users. Combining intent classifiers, chit chat, and task-oriented models that help
users achieve their goals of feeling happier while also feeling natural and conversational.
● After a few months you’re conducting deeper analysis to improve models that enable Woebot to
derive insights about individual users, thus allowing it to give personalized feedback to users, such as
“Did you realize that you’re happiest on Sundays, and least happy on Tuesdays?”
● To accomplish this you’ll work closely with our Product team to deliver these insights at the right time
and in the right manner to help users gain new insights about themselves.