Build production systems in C++ (high-performance networking/event loops), C (low-level, kernel-adjacent), and Python (ingestion/orchestration APIs).
Shipped event-driven streaming architectures (Kafka / AWS MSK) to decouple ingestion, processing, and analytics at scale.
Led an internet-scale collection platform scanning ~4.4B IPv4 per cycle, ingesting 1.5TB+/day of telemetry.
Improved ingestion reliability with idempotency + retry isolation + buffering, cutting failures from ~30% → <5%.
Built distributed observability (tracing + metrics + logs) enabling end-to-end latency diagnosis across microservices.
Stack: C/C++17, Python, Linux, Kafka/MSK, AWS (EC2/ECS/S3), Postgres, Elasticsearch, Zipkin/X-Ray
Linux staging driver cleanup (rtl8723bs) — patch series improving style/readability in drivers/staging/rtl8723bs. [C, kernel]
RTC subsystem study — sysfs ABI, wake alarms, driver registration flow, user-space tests. [kernel/RTC]
Internet-wide scanning platform — distributed collectors scanning ~4.4B IPv4; ingestion 1.5TB+/day. [Python, Kafka, AWS]
Global honeypots network — 500+ nodes (low/med/high interaction) producing detection feeds. [Terraform, Suricata]
Threat actor attribution — clustering by infra/TTPs with PassiveDNS + OSINT enrichment. [Python]
Lock-free ring buffer — bounded queue using C11 atomics (SPSC/MPMC) with microbenchmarks. [C11]
C++ distributed tracing client — trace context propagation over TCP/UDP/HTTP/custom protocols. [C++17, Zipkin/Jaeger]
Analytics dashboard backend — real-time ingestion, rolling aggregates, RBAC APIs. [Django, Postgres, Redis]
Breach analysis platform — crawler + reviewer workflows + audit trails + exports. [Django, Celery]
Linux fleet management — safe batching for SSH actions + health checks across 500+ hosts. [Django, Ansible]
Domain squatting detection — CT logs + DNS telemetry with automated takedown workflows. [Python, Kafka]
TOR exit relay monitoring — opt-in traffic observation to surface org-level risk signals. [Suricata, Zeek]
Master’s thesis — stacked denoising autoencoder for real-estate price prediction. [Python, TensorFlow]