Skip to content

Projects, Version 2

Each project from the program rebuilt with a deliberately different architecture, to show that the choice of model is a design decision rather than a default. Where version 1 reaches for a familiar baseline, version 2 picks a different paradigm and explains the trade-off.

Run these as Python packages rather than notebooks:

git clone https://github.com/sameerkarur/Data_science.git
cd Data_science
pip install -r requirements.txt
python projects_version2/run_all_v2_projects.py

Python Expense Tracker V2

Next-Generation Python Architecture

Browse the code

Python Task Manager V2

Next-Generation Python Architecture

Browse the code

Ds Sales Analysis Cohort Rfm V2

Next-Generation Applied Data Science Architecture

Browse the code

Ds Marketing Campaign Uplift Attribution V2

Next-Generation Marketing Analytics Architecture

Browse the code

ML Spotify Cohorts Hdbscan Pca V2

Next-Generation Machine Learning Architecture

Browse the code

ML Employee Attrition Explainable Boosting V2

Next-Generation Machine Learning Architecture

Browse the code

DL Lending Club Tabular Resnet V2

Next-Generation Deep Learning Architecture

Browse the code

DL Home Loan Risk Tabnet V2

Next-Generation Deep Learning Architecture

Browse the code

Genai Storytelling Multiagent Stategraph V2

Next-Generation Generative AI Architecture

Browse the code

Genai Virtual Pmo Swarm Simulation V2

Next-Generation Generative AI Architecture

Browse the code

Advgenai Hybrid Dense Sparse RAG V2

Next-Generation Advanced Generative AI Architecture

Browse the code

Advgenai Multimodal Creative Studio V2

Next-Generation Advanced Generative AI Architecture

Browse the code

Capstone1 Autonomous Perception Vit Efficientnet V2

Next-Generation Computer Vision & Safety Architecture

Browse the code

Capstone2 Hierarchical Probabilistic Demand Forecasting V2

Next-Generation Time-Series Forecasting Architecture

Browse the code

Capstone3 Dual Vision Bpr Recommender V2

Next-Generation Deep Vision & Tourism Recommender Architecture

Browse the code