How we route around censorship with ML-driven path selection, traffic morphing, and 142 entry-node IPs.
CensorshipVPNML routing
Writing
Long-form, light on hype. Architectures, trade-offs, and post-mortems from work on censorship measurement, OSINT pipelines, election analysis, and post-quantum communications.
How we route around censorship with ML-driven path selection, traffic morphing, and 142 entry-node IPs.
Architecture of a real-time NLP pipeline: TensorFlow models, sub-2-second latency, multi-language sentiment + entity recognition.
Benford’s Law, turnout modeling, ARIMA time-series — surfacing anomalies worth a second look.
Apache Kafka, TimescaleDB, 80 GPU NLP workers, MinHash deduplication. The pipeline behind 58M posts/day.