👋 Hey there! I'm Sanjeev A. ⚡
🌱 I’m an avid programmer from India with a passion for problem-solving, logic, and mathematics.
🌐 Fascinated by networks — I’ve worked on Beyond-5G and O-RAN technologies.
🐧 Deeply interested in Linux systems and always diving deeper into kernel-level development.
🎯 Constantly learning, experimenting, and exploring the internals of how things work.
😎 Fun fact: I actually built a keylogger just to understand system-level input handling!
INDRA-5G is a modular framework designed to optimize PRB allocation and classify traffic in a simulated B5G O-RAN setup.
Core Components:
- 🔍 Traffic Classification Module – Classifies eMBB, URLLC, mMTC using SVM, CNN, Transformer, and CapsNet models.
- 🧠 Resource Allocation Module – Allocates PRBs using:
- Static rule-based logic
- Genetic Algorithm
- Random Forest
- Deep Q-Network (DQN)
- 📊 KPI Collection & Preprocessing – Extracts real-time KPIs from Open5GS, srsRAN, and srsUE setups.
- 🎮 Simulation – Containerized and automated via Docker + Kubernetes for scalable emulation of 5G networks.
🔗 Repo: CIP-INDRA5G
A knowledge-based search assistant that leverages Graph Theory + LLMs for structured query resolution.
Core Features:
- 🧾 Extracts key phrases, builds dynamic dependency graphs
- 🤖 Uses OpenAI embeddings + vector stores to answer contextually rich queries
- 🔄 Retrieves citations, sources, and builds final documents using GPT agents
- ⚙️ CLI + Web-ready modular design for integration with any research corpus
A Trie-based agentic email automation pipeline with smart intent classification and dynamic action planning.
Key Features:
- 📧 Reads, parses, and classifies emails into categories (Business, Ads, Study, Spam, etc.)
- 🧠 Uses Trie + Rule Trees for command prediction and decision routing
- 🔁 Dynamically chooses the best agent action (e.g., reply, delete, archive, follow-up)
- 🔒 Designed with modularity and privacy-respecting logic
📞 Phone Number: 9444452444
|
|
|