In this episode of the Energy Tech Podcast, Mike Flores sits down with Austin Blake — Director of Data at CNX Resources — for one of our deepest, most comprehensive conversations yet on data quality, machine learning, edge computing, anomaly detection, and what real digital transformation looks like inside a major upstream operator. This conversation pulls back the curtain on how CNX is rebuilding its entire data foundation, streaming real-time operational data from the field, and applying advanced machine learning models to critical equipment like fracture fleets, flowback operations, and multi-stage compressor stations.

WHAT YOU’LL LEARN IN THIS EPISODE

1. Modernizing Oilfield Data

How CNX moved from siloed spreadsheets to a governed Snowflake data lake with clean, consistent, ML-ready data.

2. Real-Time Edge Visibility

How MQTT + AWS IoT Core deliver sub-second frac, flowback, and drilling data—and how offset-well monitoring helps prevent frac hits.

3. Machine Learning That Works

How CNX built practical anomaly detection models using 1-second SCADA data to improve compressor uptime and maintenance planning.

4. AI vs. Reality

Why GenAI won’t fix bad data—and why traditional time-series ML still delivers the most value in oil & gas operations.

5. Austin’s Career Path

How a completions engineer became CNX’s Director of Data and what he’s learned about failing fast and delivering real value.

Chapters

00:00 – Welcome to the Energy Tech Podcast

00:16 – Introducing Austin Blake (Director of Data, CNX)

00:48 – CNX’s digital transformation and Austin’s role

01:30 – Fixing siloed data with a modern Snowflake architecture

03:05 – Data governance, medallion layers, and eliminating Excel chaos

04:05 – Austin’s background: field engineering → completions → data

06:00 – Why completions lacked real-time data historically

07:03 – Rise of Python, analytics, and why engineers must learn data

08:30 – When CNX realized it needed to fix data before ML

12:00 – Why ML fails without clean data

13:30 – The well-naming problem in oil & gas

15:00 – Misconceptions about AI “orchestrating” bad data

16:02 – Aligning drilling, completions, production & midstream around data

17:10 – CNX’s edge journey: Red Lion, vendors, and protocol challenges

18:30 – Building the edge → AWS IoT Core → MQTT pipeline

20:04 – Why sandbox environments accelerate innovation

21:30 – Hot vs cold data paths: real-time vs long-term reporting

24:00 – Real example: completions dashboards in Grafana

25:20 – Preventing frack hits with offset-well observability

28:00 – Demo walkthrough: real-time frack visualization

32:50 – Why smaller operators can do this themselves

34:00 – Building context at the edge (MQTT/Sparkplug)

35:20 – Machine learning on compression fleets

37:30 – Using 1-second SCADA for anomaly detection

39:50 – Deploying ML models + measuring success via uptime

42:00 – Austin’s personal approach: iterate fast, fail fast, deliver value

44:30 – AI vs ML: what matters in oil & gas

47:00 – Closing thoughts + subscribe

Music: Uygar Duzgun / “Fast Life” / courtesy of www.epidemicsound.com