Skip to content
View preetamvr1595's full-sized avatar

Block or report preetamvr1595

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
preetamvr1595/README.md
Preetham V R — AI and Machine Learning Portfolio Animated introduction
GitHub LinkedIn Projects
Portfolio Gmail Profile views



I don't just want to understand AI. I want to build useful things with it.

An MCA student focused on AI & Data Analytics, developing practical skills in machine learning, data science, and AI-powered applications.


01 — The person behind the code

Developer working at a computer

Hi, I'm Preetham V R — an aspiring AI/ML Engineer and Applied AI Engineer from India.

I'm building my foundation through hands-on projects, experiments, and practical implementation. My goal is to understand not only how a model works, but also how to turn data and algorithms into applications that people can use.

  • 🎓 Education: MCA — AI & Data Analytics
  • 🧠 Focus: Machine Learning, Data Science, Applied AI
  • 🔬 Learning by building: Experiments, model comparisons, and end-to-end applications
  • 🛠️ Current approach: Understand the concept → implement it → evaluate it → document it
  • 🎯 Career goal: AI/ML Engineer or Applied AI Engineer

My engineering philosophy: Build with curiosity. Measure with evidence. Improve with feedback.



02 — Inside my AI systems lab

I organize my work around three connected areas.

🧠 Machine Learning

Learning how data becomes predictions.

  • Data cleaning and EDA
  • Feature engineering
  • Classification and regression
  • Model comparison and evaluation
  • Experiment-driven learning

🤖 Applied AI

Exploring how intelligent capabilities become usable tools.

  • AI-powered analytics
  • Natural-language data interaction
  • Text classification
  • Practical application workflows
  • Human-centered AI interfaces

📊 Data & Engineering

Building the foundations behind reliable results.

  • Python and NumPy
  • Pandas-based analysis
  • Data visualization
  • APIs and application structure
  • Reproducible experiments

03 — Featured projects

These are the projects that best represent the areas I am exploring. Each repository is a chance to practice a different part of the AI and software development workflow.

🔷 01. InsightAI Analytics

AI-powered data analytics platform to explore datasets, spot patterns and interact with data

Stack: React FastAPI Python Pandas NumPy Scikit-Learn Recharts

  • CSV & Excel analysis, statistical summaries, data-quality insights
  • Interactive visualizations & correlation analysis
  • ML-assisted analysis & natural-language dataset interaction
  • API-driven full-stack architecture
Source repository

🏡 02. Bengaluru Real Estate Analytics & ML

Turning property data into insights

Stack: Python Data Analysis Feature-Engineering Regression Model-Evaluation

  • Which property features matter most?
  • How to clean & prepare the data?
  • Which model performs best on unseen data?
  • How to understand & communicate prediction errors?
Real estate project repository

💬 03. Sentiment Analysis

Teaching machines to interpret text: classical ML vs neural networks on 50,000 IMDB reviews

Stack: Python Scikit-Learn NLP TF-IDF PyTorch

  • Text preprocessing & exploratory analysis
  • TF-IDF features with Logistic Regression & Linear SVM
  • Neural approaches with PyTorch, plus model comparison
  • Best: TF-IDF word pairs + Linear SVM at 91.7% accuracy
Sentiment analysis repository

🩺 04. Diabetic Retinopathy Project

Exploring AI for medical-image analysis

Stack: Medical-Imaging Preprocessing Model-Evaluation

  • Dataset & image preprocessing
  • Model architecture & training approach
  • Evaluation results & class-wise performance
  • Limitations, potential bias & responsible use
  • ⚠️ Educational project, not a medical diagnostic tool
Diabetic retinopathy repository

⚡ 05. NumPy Mastery

Strong foundations for numerical computing, through exercises & notebooks

Stack: NumPy Python Jupyter

  • Arrays, indexing & slicing
  • Broadcasting & vectorization
  • Reshaping & array operations
  • Numerical statistics & linear algebra
  • Practical numerical-computing experiments
NumPy mastery repository

🎓 06. Student Dropout Early-Warning System

Flagging at-risk students from first-semester data (4,424 students, 35 features)

Stack: Python LightGBM XGBoost MLflow SHAP FastAPI Docker GitHub-Actions

  • Leakage-free features; LightGBM & XGBoost tuned with cross-validation
  • Cost-based F2 threshold: 0.91 ROC-AUC, 90% of dropouts caught
  • MLflow experiment tracking & SHAP explanations
  • FastAPI service (single & batch, top risk factors per student)
  • 7 automated tests, CI, Docker & AWS Lambda configs
Dropout early-warning repository

🧬 07. Custom Code LLM (in progress)

Fine-tuning a 7B base model step by step

Stack: Python PyTorch Hugging-Face LoRA DPO Kaggle

  • Five sequential LoRA phases: foundation, coding, debugging, diversity, style
  • DPO on correct-vs-rejected code pairs
  • Triple quality gates (unit tests, LLM-judge score)
  • Sequence packing for 2x+ faster training
  • 400K+ training examples on Kaggle GPUs

🚀 More on GitHub

Every experiment, notebook and project in one place

Stack: Python Machine-Learning Data-Science

  • Practical ML, NLP & analytics projects
  • Ongoing experiments & learning notes
  • New projects added as they ship
All repositories

📈 Results snapshot

Project What was measured Result
🎓 Dropout Early-Warning ROC-AUC on held-out test set 0.91
🎓 Dropout Early-Warning Dropouts caught (recall) 90%
💬 IMDB Sentiment (50,000 reviews) Best model: TF-IDF word pairs + Linear SVM 91.7%
💬 IMDB Sentiment Three PyTorch networks (for comparison) 87.6 – 89.1%
🧬 Custom Code LLM Training examples across 5 phases 400K+



04 — My technology toolkit

These are the tools I use or explore across my current projects. My proficiency is developing through practice.

Languages & Data

Python, Java, and MySQL

Machine Learning & Analytics

NumPy Pandas Scikit-learn PyTorch Matplotlib Seaborn

Application Development

React, FastAPI, Git, GitHub, and VS Code

Working Environment

Jupyter REST APIs

Databases

MongoDB, PostgreSQL, Supabase

Web Development

HTML, CSS, JavaScript, TypeScript, Next.js, Vite, Flask

Gen AI & Analytics

Hugging Face XGBoost Kaggle Tableau Power BI Excel

ML Pipeline: Tools at Every Stage

🧱 Stage 🔧 Tools I use
Data & EDA Python · NumPy · Pandas · Matplotlib · Seaborn
Features & Models Scikit-learn · XGBoost · PyTorch
Evaluation Cross-validation · ROC-AUC · F2 threshold tuning
LLMs & RAG Hugging Face · ChromaDB · LoRA · DPO


05 — What I'm learning next

My roadmap is focused on building depth, not collecting technology names.

🔬 Machine Learning

  • Build complete Scikit-learn workflows
  • Understand cross-validation and data leakage
  • Compare models with appropriate metrics
  • Improve feature engineering
  • Perform error analysis and model interpretation

🚀 Applied AI Engineering

  • Improve API design and error handling
  • Add automated tests to projects
  • Learn model serving and deployment
  • Explore LLM applications and retrieval systems
  • Build reproducible, documented projects

06 — GitHub activity

GitHub Stats    Top Languages
GitHub contribution streak

07 — Let's connect

I'm interested in learning from engineers, collaborating on practical projects, and growing through feedback.

If you're working on machine learning, data science, or useful AI applications, I'd be happy to connect.

Connect on LinkedIn Visit my portfolio Email Explore my repositories



Contribution snake eating my commits

Curiosity starts the experiment. Engineering turns it into something useful.

Decorative footer

Pinned Loading

  1. Cyber_Plagarism Cyber_Plagarism Public

    Python

  2. numpy-100 numpy-100 Public

    Forked from rougier/numpy-100

    100 numpy exercises (with solutions)

    Python 1

  3. New_Diabetic_Retinopathy_Project New_Diabetic_Retinopathy_Project Public

    JavaScript

  4. whatsapp-crm whatsapp-crm Public

    Self-hostable WhatsApp CRM built with Next.js 16, Supabase, and Meta Cloud API — featuring a shared team inbox, contact CDP, sales pipelines, mass broadcasts, no-code automations, and interactive W…

    TypeScript 1

  5. ai-housing-project ai-housing-project Public

    JavaScript

  6. ai_insignt ai_insignt Public

    ai

    TypeScript