dvara
High-speed malicious URL detection using a probabilistic Bloom Filter pipeline. Precision security for technical architectures.
Two-Stage Detection Pipeline
Zero False Negatives
Guaranteed identification for all indexed malicious signatures. Always caught if added to the upstream dataset.
145K URLs/sec
Incredible local throughput optimized for high-traffic firewalls and security scanning pipelines.
5.14 MB Filter
Extremely compact memory footprint. 268,970 URLs compressed into a portable probabilistic bitset.
Filter Statistics
Performance Benchmarks
Generated using python -m dvara.benchmarks
| Metric Parameter | Performance Result |
|---|---|
| Local Bloom lookup latency | ~0.003ms |
| Throughput | ~145k URLs/sec |
| Indexed malicious URLs | 268,970 |
| Filter size | 5.14 MB |
| False positives | 0 / 100,000 tested |
Download & Install
terminal Download from CLI
code Quick Commands
Why 'dvara'?
"Every URL is a gateway. dvara stands at that gateway and decides what gets through."
About the Creator
Built by Dhruv, an Information Science student at BMSCE, Bangalore focused on systems design, applied ML, and infrastructure oriented engineering. dvara began as an exploration of how probabilistic data structures could be applied to real world threat detection, and evolved into a fully deployed malicious URL detection system with a production ready CLI and backend architecture.