A well-documented, tested FastAPI/Python prototype implementing capture through reconstruction, semantic segmentation, twin generation, policy training, optimization, and simulated deployment, but with minimal demonstrated adoption and several explicitly alpha/stubbed paths.

STELLA is a well-documented, tested FastAPI/ML solar-flare warning prototype with a React-facing API, Neupert-effect features, adaptive thresholds, and India infrastructure risk modeling, but it has no demonstrated adoption yet.

SPANDANA is a substantial traffic-incident decision-support platform with FastAPI, React, calibrated ML services, dispatch optimization, ingestion safeguards, and CI, but has no demonstrated adoption, tests flag, or license.

Kavalan is a substantial typed React/Vite climate-resilience prototype for Chennai, combining multilingual alerts, maps, crowdsourced reports, Supabase realtime data, weather syncing, and flood-image analysis, but has minimal adoption evidence.

A substantial malaria-cell deep-learning project with CNN training, MC-Dropout triage, Grad-CAM/SHAP explainability, notebooks, and FastAPI/Gradio interfaces, but only 1 star and no automated tests or CI.

A documented EVSE cybersecurity research project with three experimental notebooks, leakage-aware temporal validation, ensemble models, and SHAP analysis, but no visible adoption, tests, or CI.

Coursework repository containing five Jupyter assignments covering Python lists/tuples, pandas filtering, FAQ keyword search, NumPy operations, and Matplotlib plotting; it is functional but not packaged as a production project.

A documented, runnable Flower/PyTorch federated-learning prototype with synthetic non-IID EV-charging IDS data, FedAvg simulation, poisoning demonstration, and a networked client/server path, but no validation or adoption evidence.

A documented two-file Gymnasium reinforcement-learning prototype with a custom airplane boarding environment and MaskablePPO training, but no tests, CI, license, or demonstrated adoption.

A minimal LeetCode solution repository containing one documented Course Schedule problem and a concise C++ topological-sort implementation, with no visible testing, CI, licensing, or broader project scope.

IndiaREML is an early Colab-oriented research notebook for testing Ricardian equivalence, but currently has minimal documentation, no reproducibility scaffolding, and no visible adoption or sustained repository history.

A tiny system-design notes repository with a one-line README and a single load-balancer markdown lesson; it has no tests, CI, license, or implementation code.