awesome-data-leadership

Describes the virtues and challenges around achieving a customer-centric, data perspective in a business.

Wonderful discussion of the challenges of measuring a data team's impact, and provides clear examples of good, so-so, and poor metrics for measuring this performance.

Despite the rapidly-evolving/growing data stack, poor data quality remains an enormous problem; the article breaks it down into "downstream" and "upstream" categories.

A list ranging from the executive team doesn't have a clear vision for it's analytics program to nobody knows the quantitative impact that analytics is providing

A brief, digestable list of management principles for new engineering managers.

A framework for thinking throughout burnout including: 1) Define your personal range of tolerance, 2) Pick your career progression, 3) Pick your life progression.

Talks about an engineering management "event loop", where you touch base on people, projects, process, and self on daily, weekly, and monthly basis.

Provides 12 strategies managers can utilize to support their team and prevent burnout

Describes how leaders that know when, where, and how to shift gears between a top-down/take charge personas (“exercise authority” mode) and a more “flat” mode (in which the leader levels the hierarchy and shares power) will tend to be more successful, research shows.

Fantastic general engineering management book covering tooics such as career laddering, giving and receiving feedback, setting team culture, and more.

A pre-release of a book that gives a thorough accounting of the history of Data Science, a high-level understanding of its applications, and the ethical and social concerns associated with it.

A senior data scientist gives general DS career (some of which is worth noting as a leader) including topics around interviewing, productivity, communication, time estimation, and more.

Describes conditions and infrastructure needed for data scientists to thrive in an organization, and puts it in yhe context of data maturity.

Provides an overview of the philosophy behind Uber's ML education program.

The author describes a few process-based techniques for increasing ML project success (e.g. establishing project pilots and copilots, literature reviews, methods reviews, etc).

Provides suggestions on how teams can improve trust in ML in their org, including defining metrics up front, following some best practices when developing the model, A/B testing the model upon deployment, and more.

Proposes practical governance mechanisms for managing bias, fairness, and accountability across the full lifecycle of enterprise LLM systems.

A short book intended for product managers and product designers that describes the value of North Star metrics and how to iddentify them.

A short technical (but very accessible guide) to setting up a simple experimentation "platform" with elements of logging, measurement, assignment, and analysis.

The vast majority of business ideas fail to generate a positive impact, and this underscores the value of measuring impact, collecting data, and testing.

Provides a high-level overview of how to infuse decision-intelligence into an organization, along with some additional reading sources.

Outlines a thorough growth model that is broadly applicable to most B2C organizations where users subscribe to a service and includes discussion on how various "levers" of this growth model were tested.

1. What is the impact? 2. Do you have a signed spec answering the why and the what ? 3. Do you have your estimate of the cost? 4. Make the cost of dropping what you're doing very clear.

While every organization’s data platform approach will vary based on the industry and the size of their company, this quick and dirty guide lays out a blueprint for a modern data platform.

Gives high-level summary of data the several phases of data infrastructure that organizations mature through (from tiny start-up looking at manually-generated spreadsheets to more mature organizations with complex ETL DAGs.

Provides an overview on how to convert events from an event-driven microservice architecture into relational tables in a warehouse like Snowflake, the advantages of this architecture, and how you might want to structure your event messages.

Probably more relevant for CTO roles, but with interesting nuggets for Heads of Data, this post gives an overview of the various infrastructure and tools used in the modern startup (languages, infrastucture as code, secrets management, databases, etc).

The authors conducted a literature review and interviews with experts to create an aggregated overview of the necessary principles, components, and roles, as well as the associated architecture and workflows surrounding "MLOps"

Great talk outlines how DevOps principles can be applied to AI, and then shows in detail how CI/CD, version control, model storage, and more fit into a great MLOps process.

Describes a nice framework for thinking about job levels, based on scope and level of project complexity.

Describes the current state of confusion around data titles (using the "analytics engineer" as an example), and describes how the tech industry overvalues technical skills at times.

Provides suggestions on how developing junior talent: blocking off time for personal development, celebrating this blocked off time, hiring tutors, and more.

Covers how should data scientist roles be defined (analysis vs building), where should data scientists report (centralized vs decentralized), where should the data science function live (engineering org vs product org vs independent consultancy), and what should an organization do to set up data science for success.

The Silicon Valley treats engineers as autonomous adults who are smart people because that’s who they hire because that’s who can do the work they need done, while traditional companies tend to keep developers in pure execution roles.

Although not directly about data this is relevant: a framework for engineering managers to think through titles and expectations (including domains of technology, systems, people, process, and influence).

Briefly surveys the problem of poor data governance, describes an idea data governance model, and provides six ways to drive data-governance excellence.

A bit more for an academic or research audience, provides a style guide and suggestions on making an effective data dictionary and nomenclature for data models.

Proposes switching from tech- to user-centric data management by i) integrating data into company culture (raising awareness, tracking adoption); ii) making data governance options actionable for stakeholders outside of the data platform and iii) introducing ownership of tests on data quality.

Outlines an iterative framework (with examples) to introduce data governance within an organization (includes identify the chief data problem(s) to solve, set clear goals to resolve these problems, start small before you go big, drive incremental action, and then measure progress and iterate).

A conversation among many data practitioners about how their organizations handle data access control, data tagging, anonymization, and other key compliance activities, and what frameworks they have found helpful.

Briefly describes the importance of data contracts, provides an example of a complaint against contracts, and then how complaints arise because practitioners are stuck in the “data is oil” paradigm i.e. assume that the data is extracted, rather than deliberately creating data.