--- tags: [ingles, speaking, practica] created: 2026-10-02 status: en-curso --- # Video 2 — How is it going? ## Intro Hello again. This is my second practice video. My name is Julio. I'm from Lima, Peru. I live in Valencia, Spain. Today I want to tell you a story: how it's going for me. My English, my work, and something new that I'm learning. I practice with the method "English for Cholos": listen a lot, speak every day, and don't translate in your head. I will read for ten to fifteen minutes. My English is basic. Let's go. ## The story: how I got here I arrived in Spain at the end of July, in 2025. I came from Lima. Lima is big and noisy, and the food is amazing. In Valencia I work at the university. I'm a PhD student. I study satellite images. I look for clouds and methane leaks. Clouds are a problem, because they cover the ground. Methane is a gas that heats the planet, and my job is to find it from space. My big problem was English. I could read papers, with time and a dictionary. But I couldn't speak. When I talked with people, I understood some words, and then I lost the thread. The conversation died. It was frustrating. So this year I started a new method. Its name is "English for Cholos". It's a Peruvian course. The idea is simple: input and output. Listen a lot, and speak every day. That's my plan now: every day, listen and speak. I record my voice and I listen to myself. Some days it's bad. Some days it's a little better. But I don't stop. ## The new thing: deep learning This week I started something new. I'm studying deep learning. Deep learning is the technology behind AI: ChatGPT, image recognition, and many things now. The course is from Coursera, by Andrew Ng. I study at night and on the weekend. I take notes, and I'm writing a small book for myself, in English and in Spanish. It's a lot of work, but it's fun. The second lesson is: "What is a neural network?" I want to tell you the idea, in simple words. No math, I promise. A neural network learns from examples. For example: houses. I have a list of houses. For each house, I know two things: the size in square feet, and the price. The network looks at the list and learns the rule: bigger house, higher price. But a house can't have a negative price. So we bend the line. Below zero, the price is zero. The bend has a name: ReLU. The rule is easy: if the number is negative, it becomes zero. If it's positive, it stays the same. That bend is the start of everything. A straight line is boring: more size, more price, forever. The bend adds a rule. And when you join many of these little functions, the network learns curves, shapes and patterns. That's why it's "deep": many layers, one on top of another. The classic example is a handwritten number. The computer sees the number as a grid of pixels. Each pixel is a number: zero is black, one is white. The image has 784 pixels, so the network has 784 inputs. The output has 10 numbers: one score for each digit, from 0 to 9. The network looks at the scores and says: "I think this is a 3." The middle layers do the interesting part. Nobody tells them what to look for: they learn it by themselves. The first layers find small edges. The next layers join edges: a loop, a line, an arc. The last layers put the pieces together. The idea is old. It started in 1958, with a machine called the perceptron. A newspaper said it would walk, talk, and be conscious. Too much. The promise was bigger than the technology. Many years later, with more data and more computers, the idea worked. Now it's everywhere. ## Questions a common person asks me **What is a neural network, in one sentence?** It's a machine that learns from examples, instead of following rules that a person writes. **What is deep learning?** It's neural networks with many layers. "Deep" means lots of layers, and each layer learns something different. **Can you give me a simple example?** Sure. You upload a photo to Instagram, and it knows the faces of your friends. That's a neural network. Or a bank: it sees a strange payment and it calls you. That's also a neural network. **What did you learn today?** I learned about the bend, the ReLU. And I learned about the perceptron, from 1958. The lesson there is: big promises, small bricks. The bricks stay. **Why do you study this?** For my PhD. I want to use neural networks to find clouds and methane in satellite images. Also, I want to teach it. I'm making videos and notes about the course, in simple Spanish. Maybe they help another person like me. **Is it hard?** Yes and no. The math is hard. But the ideas are simple: examples, mistakes, and corrections, again and again. I go step by step. One small brick per day. **How is your English going?** Better than in August, but still basic. I understand more when I listen, and I can talk for two minutes without stopping. My grammar is bad, I know. But I don't stop. That's the method. **Do you get nervous when you record?** Yes, a little. The first minute is the worst. But after ten videos, I think it will be normal. Maybe. **Do you like living in Spain?** Yes, I like it. The food, the people, the weather. In Valencia, the light is amazing. But I miss Peru: my family, my friends, and the food. Peruvian food is the best in the world. Sorry, Spain. **What do you do to relax?** I go to the gym. I lift weights and I run. I want endurance, to run for more time. Some day, I want to practice martial arts too. **What is your plan for next year?** Finish the first half of the PhD. Improve my English. Learn deep learning well. And keep making videos. That's enough. ## Closing So this is how it's going. I arrived from Lima with bad English, and now I study, I record, and I learn something new every day. Some days are hard. But I'm here, and I keep going. That's all for today. Thank you for listening. If you want, leave a message and tell me your story too. See you in the next video.