AI In The Energy Sector With Wish Bakshi

Описание: In this episode, Kevin Evans chats with AI systems engineer and energy sector expert Wish Bakshi about the intersection of AI, energy infrastructure, commodity trading, and the future of open source models.

Discover how advanced AI models are transforming critical industries from oil & gas to power grids, and learn practical frameworks for deploying AI responsibly and effectively.

Main Topics:
The entire energy value chain: molecules, electrons, and bits
How AI models influence oil, gas, power grids, and commodities trading
The role of open source AI models and the lack of standardized governance
Building AI systems: from ML Ops to multi-agent orchestration
Future of AI adoption and democratization in industry

Timestamps:
00:00 - Introduction and guest overview
00:21 - Wish's background in AI for the energy sector
00:42 - Touching oil & gas across upstream, midstream, downstream
01:11 - Molecules to electrons to bits: the value chain
01:54 - Challenges of AI in production-grade energy systems
02:24 - Oil production and pipeline logistics in North America
03:30 - Use case: AI optimizing pipeline pressure and friction reduction
04:07 - Heavy crude, sulfur content, and pipeline specifications
04:49 - Venezuela’s oil reserves and global impact
05:20 - Venezuela’s current production capacity and political landscape
06:24 - Canadian oil sands as a reliable, steady source
07:14 - US oil production decline and strategic reserves (SPR)
08:36 - Data centers and energy infrastructure: scale and data needs
09:36 - Market fluctuations and AI’s role in commodities trading models
10:04 - Real-time trading models and AI adaptation to geopolitical events
11:49 - Model drift, ensemble methods, and quick adaptation strategies
12:05 - Impact of AI on stock, equity, and commodity markets
13:37 - Fundamentals vs. quant models in trading
14:07 - Incorporation of AI and quant models for profit & validation
15:03 - Types of quant models: Bayesian, probabilistic, classical ML
16:56 - Specificity of AI models in trading industries vs. general-purpose LLMs
17:04 - Leveraging LLMs and AI agents specifically for commodities and trading
18:11 - Power grid stability, the importance of the 60Hz heartbeat
19:56 - Real-time demand management and the importance of rapid AI response
20:25 - Data centers in small towns and their energy demands
22:35 - Building large AI data centers: scale and infrastructure needs
23:11 - Upgrading the North American electrical grid with industry aid
24:13 - ROI of AI in energy infrastructure and open source models
24:51 - The era of abundant creativity and rapid execution enabled by LLMs
25:21 - A day in the life of an AI systems engineer and ML Ops
26:35 - Hardware management, open source models, and cost optimization
27:05 - Managing proprietary & open source AI models on-premises
28:20 - The rise of open source models and community-driven innovation
29:01 - Licensing, weights, and the open source ecosystem
30:17 - Standardization challenges for open source AI models
31:03 - Trusted repositories like Hugging Face & community practices
32:23 - Governance, licensing, and the future of AI model regulation
33:04 - Frameworks for selecting and deploying AI models based on use case
35:04 - Orchestrator approach: observe, decide, act for multi-agent systems
36:17 - The importance of frameworks, patience, and iterative testing
37:04 - Common pitfalls: one-agent solutions, lack of planning
37:58 - Leveraging multi-agent orchestration for complex workflows
39:07 - Rapid improvements in AI accuracy over the last few years
39:24 - The accelerating adoption curve of AI technologies
41:04 - Democratization of AI tools and citizen developers
43:56 - Advice: Jump in, try all models, and learn your patterns
45:34 - The value of experimentation and cross-model comparison
46:38 - Connecting with Wish: LinkedIn & "Wednesdays with Wish" newsletter
47:19 - Wrap-up & upcoming Code to Cloud Summit
48:00 - Final thoughts & thank you

Resources & Links:
Wednesdays with Wish (LinkedIn Blog & Updates)
  / wishbakshi