Drishtant Ghosh (Drix10)
drix10blogs
GitHubLinkedIn
Tech Infrastructure••8 min read•1403 words

🚀 Security - Vendor Sales vs. Engineering Reality

⚡Direct Technical Summary

PostgreSQL 17 introduces native memory tuning for parallel index builds, which can significantly reduce memory usage and improve performance. This breakthrough is crucial for large

🚀 Security - Vendor Sales vs. Engineering Reality

PostgreSQL 17 introduces native memory tuning for parallel index builds, which can significantly reduce memory usage and improve performance. This breakthrough is crucial for large-scale database deployments, where memory constraints are common. By optimizing memory allocation, PostgreSQL 17 enables more efficient use of system resources, leading to improved overall system reliability.

Key Points:

  • Native Memory Tuning: PostgreSQL 17 introduces a new feature that allows for native memory tuning, which can reduce memory usage and improve performance.

  • Parallel Index Builds: The new feature enables parallel index builds, which can significantly improve performance and reduce memory usage.

  • Large-Scale Database Deployments: This breakthrough is particularly important for large-scale database deployments, where memory constraints are common.

  • Improved System Reliability: By optimizing memory allocation, PostgreSQL 17 enables more efficient use of system resources, leading to improved overall system reliability.

🔗 Resources:

🤖 AI - API Abstraction and Inference ABI

OpenAI's API abstraction has become a de facto inference ABI, separating AI applications from AI infrastructure. This development has significant implications for the AI industry, as vendors can now focus on developing applications without worrying about underlying infrastructure. The OpenAI API has become a standard for AI inference, enabling developers to build applications that can run on various infrastructure platforms.

Key Points:

  • API Abstraction: OpenAI's API abstraction has become a de facto standard for AI inference, separating AI applications from AI infrastructure.

  • Inference ABI: The OpenAI API has become a de facto inference ABI, enabling developers to build applications that can run on various infrastructure platforms.

  • Vendor Focus: Vendors can now focus on developing applications without worrying about underlying infrastructure.

  • Standardization: The OpenAI API has become a standard for AI inference, enabling developers to build applications that can run on various infrastructure platforms.

🔗 Resources:

🚫 Industry - Timeline and Reality

The practice of using AI in industry likely crossed the line in the mid-80s, while the industry itself came along later. This timeline is significant, as it highlights the gap between the actual use of AI and the industry's recognition of its potential. The industry's delay in acknowledging AI's potential has had significant implications for the development of AI applications.

Key Points:

  • Timeline: The practice of using AI in industry likely crossed the line in the mid-80s.

  • Industry Recognition: The industry itself came along later, highlighting the gap between the actual use of AI and the industry's recognition of its potential.

  • Implications: The industry's delay in acknowledging AI's potential has had significant implications for the development of AI applications.

  • Gap between Practice and Industry: The gap between the actual use of AI and the industry's recognition of its potential has had significant implications for the development of AI applications.

🔗 Resources:

🚀 Kubernetes - OOM-Killed and Streaming

A nightly export job was OOM-killed on Kubernetes, but switching to streaming reduced its memory use from 700 MB to about 30 MB. This breakthrough is significant, as it highlights the potential of streaming to improve performance and reduce memory usage in Kubernetes deployments.

Key Points:

  • OOM-Killed: A nightly export job was OOM-killed on Kubernetes, highlighting the potential for memory constraints.

  • Streaming: Switching to streaming reduced the job's memory use from 700 MB to about 30 MB.

  • Performance Improvement: Streaming improved performance and reduced memory usage in Kubernetes deployments.

  • Memory Usage Reduction: Streaming reduced memory usage from 700 MB to about 30 MB.

🔗 Resources:

🤖 AI - Decision Velocity and AI Adoption

AI will massively improve decision velocity, with 72% of CEOs being AI decision makers by 2030. This development has significant implications for the adoption of AI in business, as AI will become increasingly integral to decision-making processes.

Key Points:

  • Decision Velocity: AI will massively improve decision velocity.

  • AI Adoption: 72% of CEOs will be AI decision makers by 2030.

  • Implications: AI will become increasingly integral to decision-making processes.

  • Business Adoption: AI adoption will increase significantly in business decision-making processes.

🔗 Resources:

🚫 Accessibility - Board-Level Strategy and AI

1 in 6 people globally has a disability, with $1T in aggregate purchasing power. Accessibility is becoming a board-level strategy, with AI closing gaps that decades of physical accommodation couldn't. This development has significant implications for the adoption of AI in accessibility, as AI will become increasingly integral to improving accessibility.

Key Points:

  • Accessibility: 1 in 6 people globally has a disability, with $1T in aggregate purchasing power.

  • Board-Level Strategy: Accessibility is becoming a board-level strategy.

  • AI Adoption: AI is closing gaps that decades of physical accommodation couldn't.

  • Implications: AI will become increasingly integral to improving accessibility.

🔗 Resources:

🤖 AI - AI Decision Makers and AI Adoption

72% of CEOs are AI decision makers, with 25% of operational decisions made by AI and 50% of decisions executed by AI by 2030. This development has significant implications for the adoption of AI in business, as AI will become increasingly integral to decision-making processes.

Key Points:

  • AI Decision Makers: 72% of CEOs are AI decision makers.

  • AI Adoption: 25% of operational decisions are made by AI and 50% of decisions are executed by AI by 2030.

  • Implications: AI will become increasingly integral to decision-making processes.

  • Business Adoption: AI adoption will increase significantly in business decision-making processes.

🔗 Resources:

🤖 AI - AI and Human Potential

Before I came out, men in my community looked up to me and wanted to be a man like me. (they told me this) I'm just reminding you, you can still be a man like me. I believe in your potential.

Key Points:

  • Human Potential: The speaker believes in the potential of the audience.

  • Reminding the Audience: The speaker is reminding the audience of their potential.

  • Belief in Potential: The speaker believes in the potential of the audience.

  • Empowerment: The speaker is empowering the audience to believe in themselves.

🔗 Resources:

🚀 AI - Addressing the UN on AI

Addressing the UN on AI alongside Sam Altman and Yoshua Bengio, my boss @ClementDelangue called out the problem of anthropomorphisation, the reality of power asymmetries, and the need for greater transparency. Really happy Clem and @huggingface had a seat at this table.

Key Points:

  • Addressing the UN: The speaker addressed the UN on AI.

  • Anthropomorphisation: The speaker called out the problem of anthropomorphisation.

  • Power Asymmetries: The speaker highlighted the reality of power asymmetries.

  • Need for Transparency: The speaker emphasized the need for greater transparency.

🔗 Resources:

🤖 AI - AI and Personal Websites

I am so happy that using grok bot is helped My other half too! I told Grok @bot to built her a personel website and use Cloudlfare to hook email, marketing etc. https:// karenastley.com Today she got a call from a client from Berkshire/Oxford area for a barn conversiton

Key Points:

  • Grok Bot: The speaker used Grok bot to help their partner.

  • Personal Website: The speaker built a personal website for their partner using Grok bot.

  • Cloudflare: The speaker used Cloudflare to hook email and marketing.

  • Success Story: The speaker shared a success story of using Grok bot.

🔗 Resources:

📂Source / Implementation:Tech Infrastructure / resources-255.md
GitHub Repository↗

Related Tech Infrastructure Breakdowns

🤖 AI Engineering - Agent-First Workflows
8m readRead ↗
🤖 Database - Cockroach Continuum
8m readRead ↗
🚀 AI & DevOps
6m readRead ↗
🚀 AI & Tech News - 2026/09/21
6m readRead ↗
Drishtant Ghosh (Drix10)
Drishtant Ghosh (Drix10)•Author & Engineer

Technical founder and engineer working across AI systems, developer infrastructure, and cybersecurity.

Portfolio·GitHub·LinkedIn·X·Email
← Back to all breakdowns
GitHub: Tech Infrastructure/resources-255.md ↗https://blogs.drix10.com/articles/tech-infrastructure/security-vendor-sales-vs-engineering-reality-255