587: Data Engineering for Data Scientists

Episode Summary

Executive Summary: Mark Freeman discusses his path from public health and aspiring physician to senior data scientist/data engineer at Humu, where he uses data, ML, and behavioral science to improve workplace behavior. He explains leveling in data roles, startup prioritization, SQL/R/Python tooling, hiring signals, and why blockchain/Web3 data is a compelling area for data professionals.

Main Topics: Career path from public health to data science: Mark describes how a desire for social impact, a turning point in grad school, and exposure to statistics/coding shifted him from medicine toward data science and tech startups. Content creation and LinkedIn growth: He explains how daily posting began as a layoff-response strategy, then became a compounding career engine that attracted opportunities, collaborators, and mentorship. Humu’s mission and product mechanics: Mark outlines how Humu uses behavioral science, nudges, and machine learning to improve employee experience, manager effectiveness, and workplace outcomes. Data scientist leveling and promotion: He distinguishes junior, senior, and staff data science by increasing autonomy, business impact, and cross-functional influence, using a specific promotion example at Humu. Data engineering as a data scientist: Mark explains how startup realities pushed him into data engineering: data sourcing, ETL, warehouse modeling, self-service analytics, and reliable data flows. Tooling: SQL, BigQuery, R, Python, and software practices: He emphasizes SQL and managed warehouses for speed, R for multilevel modeling and observational analysis, and Python/software engineering practices for production work. Web3/blockchain as a data opportunity: Mark argues that blockchain is a massive real-world dataset and a useful frontier for data science, describing his advisory work with a DAO and NFT/network analysis.

Key Arguments: Posting consistently on LinkedIn can create inbound opportunities and mentorship because many people observe content without engaging; visibility compounds over time. At Humu, the core challenge is matching the right behavioral intervention to the right person at the right time, which makes data science central to product value. Senior data scientists are less task-driven and more responsible for identifying high-impact projects, scoping them, and prioritizing business outcomes. Staff-level work is defined by cross-organizational influence, not just coding volume: it requires buy-in, strategy, and delegation. Data engineering is increasingly part of data science in startups because reliable data sourcing, transformations, and self-service access often matter more than the analysis itself. SQL and managed data warehouses are often the fastest path to insight at scale; knowing them well can be more valuable than relying first on Spark or Python. Observational studies and multilevel models are especially important when data scientists do not control data collection and need to handle hierarchy and bias. Software engineering best practices—Git, pull requests, unit tests, modular code, code review—make data scientists more employable and effective. Startups reward prioritization and pragmatic technical tradeoffs; building the perfect system too early can be a mistake when product-market fit is still uncertain. Blockchain/Web3 should interest data professionals because it offers large, public, novel datasets and new use cases like fraud detection and transaction network analysis.

Data Points: Podcast episode number: 587 - Episode featuring Mark Freeman on the Super Data Science Podcast. Humu funding: $100 million - The company raised venture capital to support its workplace-improvement mission. LinkedIn following: Over 19,000; approaching 20,000 - Mark’s audience has grown rapidly through weekday posting. Posting cadence: Monday through Friday - He now posts content daily on weekdays. Company size at join: Around 50 people - Mark describes Humu when he joined and how startup roles were broad. Company size later: Around 130–150 people - Shows how Humu scaled during his tenure. Early analysis estimate vs actual: 5 hours scoped, 20 hours spent - Example illustrating hidden complexity in sourcing and preparing data. Years since layoff: About 2 years ago - He says a layoff during the pandemic triggered his content strategy. Side-hustle/company count: 5 companies - He says he has started five companies, with content creation being the first to make money. Daily content strategy: 1 piece per day - Initial LinkedIn plan during his layoff period. Hierarchical scale: Thousands of people / billions of dollars - Used to describe the scale of managed services like BigQuery and why they outperform ad hoc scripts. Career progression at Humu: About 1 year as data scientist before senior promotion - He moved from data scientist to senior data scientist after demonstrating broader impact. Blockchain advisory amount: Multiple months of intermittent involvement - He describes fluctuating but ongoing engagement with Charlie DAO.

Pivotal Quotes: "I’m going to create one piece of LinkedIn content that provides value to at least one person every single day." — Mark Freeman: Explaining the origin of his LinkedIn content strategy during his layoff period. "Senior data scientists come up with their own commercially impactful work." — John Crohn: Summarizing the distinction between junior and senior levels after Mark’s explanation. "I’m literally crazy enough to think that I’m going to change the world." — Mark Freeman: Describing his mission-driven motivation and belief in scaling impact through data.

Implications: The episode argues that modern data careers are increasingly hybrid: data science, engineering, product thinking, and communication all matter. For listeners, the path to growth is visible work, business alignment, and building systems that scale impact.