The Data Praxis

Insights on Data Governance, AI Governance, AI Products, Data Architecture, and Product Strategy. By Vikas Pratap Singh.

Read the blog About me

Latest Articles

View all →

AI Products & Strategy April 19, 2026 · 17 min read

Gemma 4, Decoded: Why Google Released It Free and How It Actually Works

Google released Gemma 4 under Apache 2.0 on April 2, 2026. The license change is the real story, not the benchmarks. This is the three-tier framework for 'open' AI (closed, open-weight, open-source), a technical breakdown of how Gemma 4's MoE and multimodal pipeline work, and a practitioner decision flow for picking the right tier.

#gemma-4#open-weight#open-source#ai-strategy

AI Governance & Safety April 11, 2026 · 10 min read

Mathematical Privacy Guarantees: Differential Privacy and Synthetic Data

Part 6 showed Meridian adopting differential privacy for query analytics. This article explains why: what epsilon means, how noise is calibrated, what Apple and the Census Bureau chose, and when to use synthetic data instead.

#data-governance#data-privacy#data-protection#privacy-engineering

AI Governance & Safety April 11, 2026 · 12 min read

Privacy-Preserving Computation: Encrypted Processing, Federated Learning, and the Explainability Paradox

Part 6 showed Meridian rejecting federated learning (single-tenant architecture) and deferring homomorphic encryption (47x latency). This article explains the mechanics behind those decisions, introduces secure multi-party computation, and reveals the tension between GDPR's explainability mandate and privacy protection. Concludes with a capstone PET decision framework spanning Parts 7 through 9.

#data-governance#data-privacy#data-protection#privacy-engineering

AI Governance & Safety April 11, 2026 · 9 min read

Privacy-Enhancing Technologies: Masking, Tokenization, and De-identification

Part 3 introduced PETs as governance decisions. Part 6 showed Meridian evaluating them. This article explains how each technique actually works: static and dynamic masking, vault-based and format-preserving tokenization, and the k-anonymity family of de-identification methods.

#data-governance#data-privacy#data-protection#privacy-engineering

AI Governance & Safety April 11, 2026 · 7 min read

The Benchmark Illusion: Why Passing Safety Tests Means Almost Nothing

A study of 32 models across 56 jailbreak techniques found attack success rates jumping from 0.6% to 96.3% depending on the attack type. The Safety Tax costs 7-31% accuracy. Ten alignment techniques are competing to solve this. None covers the full attack surface.

#ai-governance#ai-safety#llm-alignment#model-risk

AI Governance & Safety April 11, 2026 · 7 min read

The $0.20 Jailbreak: Why LLM Safety Alignment Is Shallow

Fine-tuning GPT-3.5 on 10 examples for $0.20 strips its safety guardrails. Removing safety from Llama 3 takes 5 minutes on one GPU. This article explains the mechanism: safety alignment concentrates in the first few output tokens, creating a shallow defense that fine-tuning, prefilling attacks, and adversarial suffixes bypass trivially.

#ai-governance#ai-safety#llm-alignment#model-risk

Perspectives

View all →

Essays on thinking clearly, living intentionally, and building a career worth having.

Philosophy & Psychology April 12, 2026 · 9 min read

Truth Waits Patiently: Why Reality Always Collects the Cost of Denial

You can ignore truth, argue with it, or replace it with a more comforting story. None of that makes it disappear. Reality is patient. And when truth returns after a long period of avoidance, it rarely arrives as information. It arrives as correction.

#truth#self-deception#mental-models#self-awareness

Productivity & Systems April 3, 2026 · 10 min read

Deep Work in the Age of AI: Why Focus Is Your Last Unfair Advantage

Current AI can assist with routine cognitive work, but accountable judgment, creative synthesis, and sustained concentration remain human strengths. Deep work is now the scarcest, most valuable skill a knowledge worker can build.

#deep-work#focus#AI#productivity

Productivity & Systems April 2, 2026 · 13 min read

You Already Know How to Run a Good Virtual Meeting. Here Is Why You Are Not Doing It.

78% of workers say meetings prevent real work. Camera-on can increase fatigue, while camera-off can reduce how engaged you appear. The problem is not knowledge. It is that obvious best practices feel optional under pressure.

#productivity#virtual-meetings#remote-work#systems-thinking

The Data Praxis

Latest Articles

Gemma 4, Decoded: Why Google Released It Free and How It Actually Works

Mathematical Privacy Guarantees: Differential Privacy and Synthetic Data

Privacy-Preserving Computation: Encrypted Processing, Federated Learning, and the Explainability Paradox

Privacy-Enhancing Technologies: Masking, Tokenization, and De-identification

The Benchmark Illusion: Why Passing Safety Tests Means Almost Nothing

The $0.20 Jailbreak: Why LLM Safety Alignment Is Shallow

Perspectives

Truth Waits Patiently: Why Reality Always Collects the Cost of Denial

Deep Work in the Age of AI: Why Focus Is Your Last Unfair Advantage

You Already Know How to Run a Good Virtual Meeting. Here Is Why You Are Not Doing It.

Stay in the loop