It's been a while since I introduced myself. My name is Gaurav Sharma. I grew up in Rudrapur, Uttarakhand. I have participated in 2 hackathons. In one of those, my team came in the top 10 among 50 teams. I also taught students for 2 years Math and CS. I also work at The Entrepreneurship Network as a Data Science Intern. I create videos for fun, also documenting my journey in the form of videos on YouTube.

I built a 4-agent AI system that researches anything for me. The hardest part wasn't the AI. It was teaching it to criticize itself. Here's what I learned building a multi-agent research pipeline 🧵 Most people use AI like this: Ask → Get answer → Hope it's right. I wanted to know: can I build a system that checks its own work? So I built one with 4 specialized agents: 🔍 Search Agent → finds sources 📖 Reader Agent → scrapes & condenses ✍️ Writer Agent → drafts the report 🧐 Critic Agent → flags weak claims Pipeline: search → read → write → critique → revise. Biggest takeaway: Building AI agents isn't about the model. It's about designing the loop between agents. A good critic is worth more than a better writer. GitHub: github.com/gaurav-aiml/AI_Mu…

Day - 4 of completing fine tune project Today I have implemented training dataset and fine tune the qwen3:0.6B model because of hardware limitation. #ai #finetune

Day - 3 of completing fine tuning project Today I have implemented the the tokenize.py where it will take the json format data after cleaning and deduplication. It will convert to tokens for training the llm. #ai #finetuning

Day - 2 of completing the fine tuning project. Today I have implemented the clean_text.py for removing of stop words, deduplication that can be train the ollama model. Then the extracted text convert to json.

Day-1 Extracting text from pdf. Few days back I wanted to start project on fine-tuning tuning. I was confused how can I start this. When I start the project I just give the prompt and download the code file run and it failed terrible. At that point of time I came to know by just downloading code file is not the good way for creating project. So I breakdown the whole project. I am fine tuning the qwen3:0.6b on my gpu 1650 nvidia. This is the day 1 of fine tuning the qwen3:0.6b.

Covariance vs Correlation 📊 Covariance tells you how two variables move together: → Positive: same direction → Negative: opposite direction → Zero: no linear relationship But covariance depends on the variables' units, making its magnitude hard to compare. Correlation = standardized covariance It ranges from −1 to +1: → +1: perfect positive linear relationship → −1: perfect negative linear relationship → 0: no linear relationship Key idea: Covariance tells you the direction of joint variation. Correlation tells you the direction + strength, on a unit free scale. #Statistics #DataScience #MachineLearning #Python

Everyone Know about Linux. Do you Know, What actually Linux is, Why it is created. I have one Question. Is Linux is an OS or Kernel?