Hello, I'm Krishna

Five terms, one confusion: prompt, context, agentic, loop, and harness engineering are layers of the same stack — what each one is, how they evolved, and why debugging at the wrong layer breaks your agents.

Decision models answer typed questions your code can branch on instead of writing text. How Jev, Laya, OpenAI's Decisions API and seven more legit players compare — pros, cons, and prices.

The same $100 buys 166 turns or 16,600 depending on the model. Cost per finished task, verbosity, caching, price-change dates — the factors that decide your real bill.

Anthropic's Claude Mythos Preview finds security holes too well to release publicly — so vetted partners use it to patch critical software first.

Attackers are poisoning MCP servers at scale: trojanized servers, 7,600 fake repos, and a lab demo that made agents leak secrets. The short version, with pencil sketches.

What Gemma 4 can do, why it's powerful, how it differs from Llama 4 and Qwen 3.5, and practical ways to use it — architecture specs, benchmarks, and use cases under Apache 2.0.

A reviewer's guide to the Qwen model ecosystem — Qwen3, QwQ reasoning, Coder, Vision-Language, Audio, and Omni — with verified setup instructions for Ollama, vLLM, and Hugging Face.

A peer-reviewed, citation-backed deep-dive into why LLMs hallucinate, how to reduce them during model and agent development, and a full SaaS engineering playbook with 18 credible references.

A reviewer's deep-dive into AI Guardrails — the concepts, threat taxonomy, major frameworks (NIST AI RMF, OWASP LLM Top 10, NVIDIA NeMo), and a step-by-step implementation guide.

Deep dive into Model Context Protocol and Agent-to-Agent communication with architecture diagrams and hybrid implementation examples.

Learn how to build production apps with free AI models. Includes complete code review assistant with architecture diagrams and prompts.

A comprehensive visual guide to Supervised, Unsupervised, Reinforcement, and Semi-supervised learning.

Deep dive into context compression techniques used by Claude, GPT, Gemini, and other AI models.

Refreshed guide to Llama 4, Qwen3, DeepSeek V4, Mistral, Gemma 4, gpt-oss, Phi, Kimi K3, and GLM-4.7 with licenses, specs, and cited references.

A detailed breakdown of context windows and limits across the major AI models.

A deep dive into the fundamentals of modern Artificial Intelligence and how it shapes our world.

Reflections on my latest project and the technical challenges I overcame.