I am a Software Performance Engineer with extensive experience building, validating, and optimizing performance-critical systems across CPU, GPU, embedded, and Linux environments. My core strengths are modern C/C++, Python, system-level debugging, profiling, and reliability engineering.
I specialize in CPU and GPU performance analysis, including CUDA kernel optimization, memory efficiency, latency reduction, occupancy tuning, roofline analysis, and bottleneck investigation. I use reproducible benchmarks and profiling evidence to guide optimization decisions for compute-intensive workloads.
I have substantial embedded automotive experience, including ownership of BMW AUTOSAR core modules, NVRAM management, diagnostics, CAN/Ethernet communication, and integration testing. I have led small engineering teams, mentored engineers, and delivered high-quality software in complex production environments.
My validation background includes performance, power, temperature, compatibility, and telemetry analysis across Linux and Windows systems. I am experienced in analyzing performance counters, runtime metrics, logs, firmware, operating-system behavior, and hardware interactions to identify regressions and root causes.
I also work with AI coding evaluation and agentic development workflows. I design Dockerized evaluation environments, build robust verifiers, evaluate AI-generated solutions, and use prompt engineering, rubrics, benchmarks, and correctness checks to improve engineering outcomes.
I am interested in roles involving high-performance computing, GPU software, systems programming, embedded systems, autonomous systems, and hardware-integrated software. I bring a practical, quality-focused approach to developing maintainable and efficient systems.
I design evaluation environments for AI agents, including Dockerized task environments, reference solutions, sealed verifiers, and binary reward mechanisms. My work spans databases, systems, frontend development, and LLM inference tasks.
I harden verifiers against reward hacking through subprocess isolation and by ensuring ground-truth data remains inaccessible to agents. I use Python, pytest, and Docker to create reproducible and reliable evaluation workflows.
I develop and optimize custom CUDA kernels for deep-learning workloads. My optimization work includes shared-memory tiling, memory coalescing, warp-level primitives, parallel reductions, kernel fusion, occupancy optimization, and Tensor Core/WMMA techniques.
I use NVIDIA Nsight Compute, NVIDIA Nsight Systems, roofline analysis, benchmarking, and correctness validation to identify memory, compute, latency, synchronization, and occupancy bottlenecks.
I direct AI coding agents through iterative GPU optimization using prompt and context engineering, profiling evidence, benchmark feedback, rubric design, correctness validation, and architecture-specific optimization guidance.
I optimized CUDA kernels using NVIDIA Nsight Compute and NVIDIA Nsight Systems, focusing on memory bottlenecks, latency, and occupancy constraints. I applied roofline analysis to distinguish compute-bound and memory-bound workloads and drive bottleneck-focused optimization.
My kernel-level work included shared-memory tiling, memory coalescing, warp-level shuffle primitives, synchronization optimization, occupancy tuning, loop unrolling, and reduction of warp divergence.
I worked with high-performance GPU libraries and concepts including cuBLAS and CUTLASS-based GEMM kernels. I also used Docker and Kubernetes for containerization and orchestration.
I owned a BMW AUTOSAR core module responsible for centralizing diagnostic fault memory and managing non-volatile memory. I led the design, development, integration, and testing of core embedded components.
I led a small engineering team through task allocation, mentoring, technical guidance, and delivery coordination. I developed components using C, C++, Google Test, and Linux.
My work included AUTOSAR NVRAM components, CAN and Ethernet communication, and UDS diagnostic functions. The module achieved the highest code-quality rating in the team according to static analysis and software quality tooling.
I led development of a complex software communication wrapper that enabled integration across AUTOSAR and proprietary software architectures. I contributed to implementation using Java, Xtext, Xtend, and C++.
I delegated development tasks, owned testing processes, and supported automated code generation. I also mentored junior engineers in AUTOSAR implementation and engineering best practices.
I was recognized for strong project performance and considered for promotion to Software Architect.
I conducted and analyzed performance benchmarks for CPU, memory, and I/O systems on Windows and Linux environments, including Ubuntu and Red Hat. I also validated system and virtual-machine compatibility across a broad range of desktop configurations.
I performed thermal and power stress testing to analyze energy consumption under extreme conditions. I developed test plans and validation content for system performance and compatibility.
I investigated performance and compatibility issues through root-cause analysis using system telemetry, runtime behavior, and validation data.
Bachelor's degree, Electrical Engineering
Bachelor's degree in Electrical Engineering.
Specialization, Embedded Systems
Embedded Systems Specialization.
Course, CUDA Parallel Programming on NVIDIA GPUs
Completed CUDA Parallel Programming on NVIDIA GPUs course.
Course, ROS2
Completed ROS2 course.
Course, Exploratory Data Analysis for Machine Learning
Completed Exploratory Data Analysis for Machine Learning course.
This professional hasn’t added portfolio projects yet.
This professional hasn’t listed any services yet.
