Senior Backend and Performance Engineer

Optimized behavioral-prediction hot paths and reduced simulation step times by up to 60%, from approximately 2,000 ns to 800 ns on a local i5-11 machine. Implemented RCU-based lock-free caching to reduce CPU contention by up to 50% in selected scenarios.

Improved GitHub Actions build jobs from 20 to 4 minutes through compiler-artifact caching. Reduced state synchronization delays by roughly three times with custom EVM batch-processing integrations, introduced reliable Zstd-compressed integration fixtures, tuned Rayon and Tokio scheduling, and mentored engineers on architecture and low-level engineering practices.

Resolved a complex Rust/C cross-language memory-corruption issue that eliminated recurring production outages and saved an estimated 20 engineering hours previously spent on more than 30 intermittent incidents per week. Reverse engineered proprietary access-point boot processes to enable deployment of lower-cost networking hardware and reduce hardware costs by up to 60%.

Developed and deployed a custom OpenWRT-based firmware platform for centralized billable Wi-Fi services across more than 400 commercial and retail locations. Built a zero-downtime A/B over-the-air firmware update system for more than 2,500 devices and a backend API layer for centralized monitoring and remote management.

Engineered ultra-low-latency execution and market-data adapters for centralized exchanges using WebSocket, FIX 4.4, and REST APIs. Optimized throughput to process more than 5,000 to 10,000 events per second.

Architected an OpenTelemetry-based distributed tracing framework across asynchronous services, reducing time to resolve performance regressions by up to 70%. Profiled latency-critical trading workflows, removed lock contention, and reduced p99 latency by 20% to 30%.

Built high-throughput packet-processing pipelines using Rust, modern C++, DPDK, and zero-copy architectures, sustaining line-rate traffic above 10 Gbps. Improved production performance by 15% to 20% through profiling, cache-invalidation reduction, and memory-layout optimization.

Fixed optimization-breaking undefined behavior caused by C-style type punning by backporting a standards-compliant std::bit_cast polyfill for GCC 10. Modernized networking components with C++20 templates and concepts, improving matcher efficiency by up to 30% and reducing code duplication by 6% to 8%.

Built a highly parallelized counterfactual regret minimization engine and distributed backend for large-scale game-theory simulations under strict confidentiality constraints. Orchestrated cloud infrastructure for ML training workflows using Amazon APIs and automation tools.

Reduced AWS EC2 compute overhead by approximately 30% by profiling and eliminating cross-node NUMA bottlenecks in concurrent solver-training workloads. Implemented an automated batch ELT workflow backed by ClickHouse, reducing historical snapshot processing latency by 98% to 99%.

Developed dynamic pricing and automatic client-suspension functionality for the cEnergo smart-metering system, coordinating it with eight existing business-logic modules. Ported an MCU bare-metal firmware codebase to a Linux-based platform in under one month.

Reverse engineered a proprietary smart-card billing protocol by capturing USB traffic from a rebranded ActiveX control and non-PC/SC T=14 card readers. Implemented reliable remote firmware reflashing for raw MTD partitions with integrity checks and developed REST API endpoints for external clients.

Completed coursework in system analysis, control, and information processing.

Studied electric drive and automatics.

Completed coursework in calculus, linear algebra, and theoretical physics.