CSE undergrad drawn to the hard parts of backend engineering — distributed consensus, low-latency I/O, systems that hold up under pressure. I work closest to the metal when I can: lock-free queues in C++, Raft from scratch in Go, SSE over polling because polling is lazy.
Interested in fintech infrastructure, distributed systems, and MLOps tooling. Currently building microservices at Flexzistay and contributing to KServe (CNCF).
Building backend systems that hold up under pressure — high-throughput queues, low-latency pipelines, distributed coordination. And increasingly, making AI actually work in production: inference routing, agent orchestration, LLMs that do real things instead of demoing well.
The intersection is where it gets interesting — autonomous systems with financial logic, agents that don't just respond but act. That's the space I want to build in.
RiftMQ — Kafka-inspired queue in Go. Raft consensus, WAL, consistent hashing. 17,900 msg/sec, 100% delivery under 200 producers.
FIX Protocol Trading Gateway — C++20 order engine. Lock-free queues, async I/O. P50: 350µs, P99: 633µs at 250K orders/sec.
LLM Inference Router — OpenAI-compatible gateway across vLLM nodes. Circuit breakers, Prometheus/Grafana observability.
#142 · Namespace filtering for multi-tenant K8s
#161 · InferenceGraph routing for multi-model pipelines