Data Quality, AI in Baseball, ODSC AI East Slides, Ollama, and Local Agents
ODSC AI West 2026 Track Update!
ODSC AI West 2026 is bringing together the tracks shaping how AI is built, scaled, and led in 2026:
Physical AI | Agentic AI | AI Engineering & AIOps | Data Science & Machine Learning | Data & AI Infrastructure | Hands-On Training | Keynotes & Industry Leadership | AI x Leadership Summit
Register by Friday for 75% off!
ODSC x IronSpark AI Careers Survey
We want to learn how you and your peers are managing careers now that organizations of all sizes have adopted AI. Complete the survey for a chance to win prizes, such as a ticket to ODSC AI West 2026, an Amazon gift card, and more.
Why Data Quality is the Main Bottleneck Facing AI Projects
Poor data quality is a long-standing problem that constrains the accuracy, reliability, and confidence of AI systems.
AI in Baseball: How Data, Computer Vision, and Automation Are Changing the Game
Explore how AI in baseball is transforming player development, fan experience, stadium operations, officiating, and front-office decision-making across MLB.
Ten ODSC AI East 2026 Slides Every Data Scientist Should Review
These ten AI slides skip generic GenAI hype and focus on the issues practitioners are handling now: transformer fundamentals, agent interoperability, responsible design, observability, evaluation, and deployment.
What Ollama Reveals About Local AI, Agents, and Open Models
Explore how Ollama is shaping local AI, open models, agent harnesses, tool design, structured outputs, and the future of AI engineering.
ODSC Highlights
What’s New at ODSC AI West 2026: New Tracks, Speakers, and More
Explore what’s new at ODSC AI West 2026, including the AI Engineering Accelerator, new conference tracks, first-announced speakers, networking events, and more.
Leadership and Careers
Affordable Ways AI Helps Tackle Retail Distribution Challenges
As AI tools become more affordable and accessible, retailers can use them to improve forecasting, inventory management, and delivery efficiency without major up-front investments.
Agentic AI ROI: Why Enterprise Value Depends on What Happens After the Demo
Agentic AI ROI depends on more than impressive demos. Learn how production engineering, governance, identity, cost control, and workflow redesign turn AI agents into measurable business value.
Data Science & AI News
MIT: AI Job Disruption Has Not Arrived at Scale Yet
New labor market data from MIT suggests AI has not yet caused broad job losses, though entry-level roles in AI-exposed fields face growing pressure.
Robinhood Brings Agentic AI Trading to Retail Investors
Robinhood is bringing agentic AI trading and AI-powered spending tools to retail investors, raising new questions about AI and investing.
NVIDIA Plans Major Taiwan Investment as AI Supply Chain Expands
NVIDIA is planning a major investment in Taiwan as Jensen Huang calls the island the epicentre of the AI revolution, strengthening ties with TSMC.
Microsoft Report Shows Uneven AI Adoption Across the United States
Microsoft’s latest AI diffusion report shows U.S. AI adoption is growing, but growth is quite uneven when comparing urban and rural regions.
Alphabet Plans $80 Billion Stock Sale to Fund AI Infrastructure Expansion
Alphabet plans to raise $80 billion through stock sales and a Berkshire Hathaway investment to fund AI compute infrastructure.
More AI News:
IBM and Red Hat Launch Project Lightwell to Secure Open Source AI Infrastructure
NVIDIA Bets Billions on Photonics to Scale AI Infrastructure
New Report Finds AI Data Center Growth Depends on Full Semiconductor Stack
U.S. Moves to Close NVIDIA AI Chip Export Loophole For Chinese Firms Abroad
Intel Expands AI Infrastructure Strategy With Rackscale Systems and Agentic Inference
New Podcast Episode: Ollama, Local AI, and Open Agents with Parth Sareen
The episode covers local-first AI, Ollama Launch, structured outputs, tool design, verification loops, context engineering, and how coding agents are changing software engineering workflows.
Zac, this is a brilliant diagnosis of the root cause behind AI stagnation. The security vulnerabilities tied to data quality are particularly alarming—the fact that altering a mere 0.01% of a dataset's samples can fundamentally compromise a model's outputs for as little as $60 proves exactly how fragile these systems truly are.
DOCX
Your recommendation to implement automated "quality gates" to intercept bad data before it reaches the models is exactly the right instinct. However, when we transition from deploying AI as an analytical tool to deploying it as an autonomous agent executing multi-step enterprise workflows, those gates cannot live strictly within the software pipeline.
DOCX
If a model is acting on poisoned or biased data, relying on a software-based quality gate to stop a catastrophic execution is a massive liability. Software cannot reliably govern software that has fundamentally learned the wrong patterns.
This is the core architectural shift behind Veritas Core. We took the concept of the automated quality gate and moved it entirely out of the software layer, anchoring it in bare-metal physics.
By integrating physical TPM 2.0 hardware circuit breakers at the PCIe layer, Veritas Core ensures that even if an AI system is operating on flawed, incomplete, or maliciously poisoned data, it physically cannot actuate an unauthorized enterprise transaction. The hardware enforces a default-deny boundary at T=0. The model might hallucinate or attempt a skewed action due to bad data, but the physical circuit mechanically opens and drops the payload before the damage can scale across the network.
Data will always be the foundation of the model's intelligence, but hardware must be the foundation of the enterprise's execution. Excellent piece highlighting a critical vulnerability.