https://krina.me/blog/ 1% better: a 39-post learning-in-public series (Jul–Aug 2026) by Krina Vaghela about AI, writing, design and marketing. en Thu, 20 Aug 2026 12:00:00 GMT https://krina.me/blog/day-39/ https://krina.me/blog/day-39/ Thu, 20 Aug 2026 12:00:00 GMT If you want to change the station on your radio, you change the frequency. So did I. I still remember three years ago, I stood in front of a skyscraper. CareerEventsPeople https://krina.me/blog/day-38/ https://krina.me/blog/day-38/ Mon, 17 Aug 2026 12:00:00 GMT Putting myself out there. Getting comfortable being uncomfortable. Lately, I’ve been spending a lot of time around founders, co-founders, student builders, and… CareerPeople https://krina.me/blog/day-37/ https://krina.me/blog/day-37/ Fri, 14 Aug 2026 12:00:00 GMT Curiosity led me to meet Aidan Gomez, co-founder and CEO of Cohere. Two weeks ago, I was curious about a paper that helped change modern Artificial… PeopleEventsAI https://krina.me/blog/day-36/ https://krina.me/blog/day-36/ Tue, 11 Aug 2026 12:00:00 GMT What I learned about chunking strategies. Documents need to be split into smaller chunks before they can be used in RAG. AI https://krina.me/blog/day-35/ https://krina.me/blog/day-35/ Mon, 10 Aug 2026 12:00:00 GMT Here’s a simple way that can make your writing easier to follow: The F-pattern. When people visit a webpage, they usually don’t read everything from beginning… WritingDesign https://krina.me/blog/day-34/ https://krina.me/blog/day-34/ Sat, 08 Aug 2026 12:00:00 GMT I learned from Andrej Karpathy’s State of GPT talk. Pretraining → Fine-tuning → Reward Modeling → Reinforcement Learning. AI https://krina.me/blog/day-33/ https://krina.me/blog/day-33/ Fri, 07 Aug 2026 12:00:00 GMT Showed up to the Build Night TO by TechTank x PostHog at Toronto Metropolitan University. The room was filled with builders. EventsPeople https://krina.me/blog/day-32/ https://krina.me/blog/day-32/ Thu, 06 Aug 2026 12:00:00 GMT What is quantization? An LLM stores billions of numerical weights. Quantization represents them with fewer bits. AI https://krina.me/blog/day-31/ https://krina.me/blog/day-31/ Wed, 05 Aug 2026 12:00:00 GMT A prompt is simply a way of communicating with generative AI. If the instructions are vague, the output will be vague. AIWriting https://krina.me/blog/day-30/ https://krina.me/blog/day-30/ Tue, 04 Aug 2026 12:00:00 GMT Today, I learned about prompt tuning. The main model stays frozen. It does not change. In AI, this is called a frozen model. AI https://krina.me/blog/day-29/ https://krina.me/blog/day-29/ Mon, 03 Aug 2026 12:00:00 GMT Ever wondered how an LLM learns to answer questions instead of simply predicting more words? Today, I learned about Supervised Fine-Tuning (SFT). AI https://krina.me/blog/day-28/ https://krina.me/blog/day-28/ Sun, 02 Aug 2026 12:00:00 GMT What is Evaluation? Today, I learned about AI evaluation. Evaluation is the process of measuring whether an AI system performs its intended task well. AI https://krina.me/blog/day-27/ https://krina.me/blog/day-27/ Sat, 01 Aug 2026 12:00:00 GMT The main idea is simple: Don’t try to automate or optimize everything first. First, make sure the work is actually needed. Design https://krina.me/blog/day-26/ https://krina.me/blog/day-26/ Fri, 31 Jul 2026 12:00:00 GMT Can machines really think? Today, I explored a question that sounds simple but is surprisingly difficult. At first, I thought the answer might be yes. AI https://krina.me/blog/day-25/ https://krina.me/blog/day-25/ Thu, 30 Jul 2026 12:00:00 GMT Today, I learned about embeddings. At first, I thought they were just another technical AI word. AI https://krina.me/blog/day-24/ https://krina.me/blog/day-24/ Wed, 29 Jul 2026 12:00:00 GMT What is an API? Here’s what I’ve understood so far. An API lets two applications communicate with each other. Tools https://krina.me/blog/day-23/ https://krina.me/blog/day-23/ Tue, 28 Jul 2026 12:00:00 GMT Why do LLMs hallucinate? But before that... What is a hallucination? Here’s what I’ve understood so far. AI https://krina.me/blog/day-22/ https://krina.me/blog/day-22/ Mon, 27 Jul 2026 12:00:00 GMT What exactly is self-attention? Here’s what I’ve understood so far. Every word depends on other words. AI https://krina.me/blog/day-21/ https://krina.me/blog/day-21/ Sun, 26 Jul 2026 12:00:00 GMT Today, I wanted to understand a question I’ve had for a while. What exactly is a Transformer model, and why is it behind many modern AI systems? AI https://krina.me/blog/day-20/ https://krina.me/blog/day-20/ Sat, 25 Jul 2026 12:00:00 GMT Today, I found two interesting open-source projects. A skill that helps AI coding assistants give the answer first instead of long, unnecessary explanations. ToolsWritingAI https://krina.me/blog/day-19/ https://krina.me/blog/day-19/ Fri, 24 Jul 2026 12:00:00 GMT Today I learned: HTML is becoming the new Markdown. Markdown is winning because it makes writing simple: plain text, readable by anyone, easy to share. WritingToolsAI https://krina.me/blog/day-18/ https://krina.me/blog/day-18/ Thu, 23 Jul 2026 12:00:00 GMT Today, I learned the fundamentals of n8n. Here’s the simplest way I understand it: What is n8n? ToolsAI https://krina.me/blog/day-17/ https://krina.me/blog/day-17/ Wed, 22 Jul 2026 12:00:00 GMT Today, I learned what self-supervision is and why it changed AI. Here’s the simplest way to think about it: Traditional AI needs labelled data. AI https://krina.me/blog/day-16/ https://krina.me/blog/day-16/ Tue, 21 Jul 2026 12:00:00 GMT Today, I learned about LLM Twins. Here’s the simplest way to think about it: What is an LLM Twin? It’s an AI version of you. AIWriting https://krina.me/blog/day-15/ https://krina.me/blog/day-15/ Mon, 20 Jul 2026 12:00:00 GMT Today, I learned the difference between AI workflows and AI agents. Here’s the simplest way to think about it: The AI follows a fixed set of steps. AITools https://krina.me/blog/day-14/ https://krina.me/blog/day-14/ Sun, 19 Jul 2026 12:00:00 GMT Today, I learned about training, pre-training, fine-tuning, and post-training. Here’s the simplest way to think about it: The process of updating a model’s… AI https://krina.me/blog/day-13/ https://krina.me/blog/day-13/ Sat, 18 Jul 2026 12:00:00 GMT Today, I learned about prompt engineering best practices. Here’s the simplest way to think about it: Tell the model exactly what you want. Remove ambiguity. AIWriting https://krina.me/blog/day-12/ https://krina.me/blog/day-12/ Fri, 17 Jul 2026 12:00:00 GMT Today, I learned why foundation models changed the way we build AI. Here’s the simplest way to think about it: Most AI models were designed for a specific… AI https://krina.me/blog/day-11/ https://krina.me/blog/day-11/ Thu, 16 Jul 2026 12:00:00 GMT Today, I learned about the 4 Ps of Marketing. Here’s how I understand the 4 Ps: What problem are you solving? The idea should solve a real problem. Marketing https://krina.me/blog/day-10/ https://krina.me/blog/day-10/ Wed, 15 Jul 2026 12:00:00 GMT Today, I experimented with Google Jules, an autonomous AI coding agent. With Jules, your codebase keeps moving forward. ToolsAI https://krina.me/blog/day-9/ https://krina.me/blog/day-9/ Tue, 14 Jul 2026 12:00:00 GMT Today, I tried a completely new workflow: I built my Motivational Quotes app using Andrej Karpathy’s LLM Wiki technique as the primary context and persistent… ToolsAI https://krina.me/blog/day-8/ https://krina.me/blog/day-8/ Mon, 13 Jul 2026 12:00:00 GMT Today, I built a project. I’ve been looking for a better way to organize everything I’m learning about AI. ToolsAI https://krina.me/blog/day-7/ https://krina.me/blog/day-7/ Sun, 12 Jul 2026 12:00:00 GMT Today I learned about RAG (Retrieval-Augmented Generation). AI + external knowledge = better, grounded answer. AI https://krina.me/blog/day-6/ https://krina.me/blog/day-6/ Sat, 11 Jul 2026 12:00:00 GMT Today I learned how to connect an MCP tool to Cursor. Here’s my simple explanation. First, choose the MCP server you want to use. For example, GitHub. Tools https://krina.me/blog/day-5/ https://krina.me/blog/day-5/ Fri, 10 Jul 2026 12:00:00 GMT Today, I spent time understanding Model Context Protocol (MCP). A few things I learned: MCP (Model Context Protocol) is an open standard that helps AI… AITools https://krina.me/blog/day-4/ https://krina.me/blog/day-4/ Thu, 09 Jul 2026 12:00:00 GMT Today, I spent time studying Steve Jobs - Introducing the iPhone (2007) presentation. A few things I learned: Start with one clear theme. WritingMarketing https://krina.me/blog/day-3/ https://krina.me/blog/day-3/ Wed, 08 Jul 2026 12:00:00 GMT Today, I spent time understanding context engineering. Context engineering is about deciding what information an AI model should have before it responds, not… AI https://krina.me/blog/day-2/ https://krina.me/blog/day-2/ Tue, 07 Jul 2026 12:00:00 GMT Today, I spent time refining my understanding of Git/GitHub. Version control is basically a way to track changes in a project over time, so you can compare,… Tools https://krina.me/blog/day-1/ https://krina.me/blog/day-1/ Mon, 06 Jul 2026 12:00:00 GMT Today, I actually tried to learn what goes behind LLMs. A few things which I learned: These models aren’t thinking or searching; they’re predicting the next… AI