Framework for ns-3
2023 - 2025
Developed a framework to automate and standardize the simulation workflow from planning to plotting, enabling researchers to focus on core strategic planning and data analysis.
⚙️ Statistical Framework for System Level Simulator (ns-3)
Developed a statistical framework for the ns-3 simulator to facilitate the collection and analysis of simulation results.
Developed a containerized statistical framework that automates the entire pipeline from parameter input to result analysis using Django and Apache Airflow. I implemented a robust data architecture using PostgreSQL and MongoDB to ensure data integrity and scalability. By leveraging Docker for the entire ecosystem, I achieved environment consistency and streamlined the deployment process.
-
Automated Workflow: Replaced manual simulation execution with Airflow DAGs, reducing human error.
-
Hybrid Database Strategy: Optimized data storage by using PostgreSQL for structured logs and MongoDB for large-scale simulation results.
-
Scalable Architecture: Ensured seamless scalability and environment parity across different systems via Docker containers.
| Category | Technologies Used |
|---|---|
| Backend & Orchestration | Django, Apache Airflow, Python |
| Database | PostgreSQL (Relational), MongoDB (NoSQL) |
| Infrastructure | Docker, Docker-compose |
| Tools & Monitoring | MongoDB Compass, Microsoft Teams Integration |
⚙️ Deep Learning Framework for System Level Simulator (ns-3)
Developed an advanced deep learning framework by integrating MLflow into the existing statistical architecture to automate experiment tracking and model management.
I established a robust pipeline between localhost and remote hosts to log hyperparameters, metrics, and artifacts in real-time. This enhancement significantly improved reproducibility and research efficiency for complex system-level simulation-based AI studies.
-
Enhanced Experiment Traceability: Leveraged MLflow Tracking Server to monitor all DL training sessions and hyperparameter tuning.
-
Scalable Artifact Management: Implemented an Artifact Proxy to securely store and retrieve large-scale model files and configurations across remote hosts.
-
Integrated Data Pipeline: Combined PostgreSQL for metadata tracking and local/remote file systems for comprehensive storage.