I find out why distributed systems are slow.

I'm a computer science graduate focused on distributed systems and high-performance computing. I'm based in Portland, Maine, and looking for engineering roles in aerospace, distributed systems, and performance-critical software in the greater Boston area.

Selected work

Two projects where I measured first and optimized second.

Distributed fine-tuning of an 8-billion-parameter transformer

I profiled and optimized distributed training on dual NVIDIA A100 GPUs, focusing on communication overhead, memory fragmentation, and scaling limits under heavily constrained hardware.

What I did

  • Analyzed communication overhead for ZeRO-2 and ZeRO-3 optimizer stages
  • Identified scaling bottlenecks through NCCL trace analysis
  • Quantified how memory fragmentation limits batch size
  • Improved total GPU utilization through configuration tuning

Result

Showed the tradeoffs between parameter sharding and throughput, and presented the findings to the University of Basel HPC research group.

PythonPyTorchDeepSpeedCUDANCCLSlurmA100

Shortest-path algorithms on social and grid graphs

I compared breadth-first search, Dijkstra's algorithm, and A* with classical and embedding-based heuristics. I built a benchmarking framework in C# to study how graph topology, density, and heuristic quality affect search performance.

What I found

  • Graph structure strongly shapes how much of the graph each algorithm explores, and how long it runs
  • Heuristic quality largely determines A* efficiency on grid graphs
C#SNAP datasetsCustom benchmark framework

Systems foundations

Implementation-focused coursework spanning parallel performance, distributed coordination, and fault tolerance.

Parallel and performance engineering

  • Cache-optimized matrix multiplication in C
  • OpenMP and MPI parallel programs
  • CUDA kernels that accelerate a heat diffusion simulation
  • Score-P profiling to guide refactoring
  • Load-balancing experiments for sparse workloads

Focus: scalability, memory locality, imbalance detection, GPU utilization

Distributed systems from first principles

  • Chord distributed hash table in Java, with node join, leave, and stabilization
  • Distributed transaction coordinator over JDBC
  • Vector clocks for causal ordering across nodes
  • Homomorphic cryptosystem with correctness tests

Focus: correctness, consistency, coordination, fault tolerance

These were assigned coursework for a graduate computer science program. The repositories are private to avoid enabling academic dishonesty. I'm happy to share access on request.

Skills

Languages

CC++JavaPythonC#

Parallel programming and systems

OpenMPMPICUDANCCLSlurmDeepSpeed

Concepts

Distributed systemsHigh-performance computingScalabilityProfilingFault tolerance

What drives me

Understanding complex systems

I'm fascinated by the engineering challenges that appear when software has to perform reliably under real constraints. I'm particularly interested in how architecture and resource management shape performance, and in the profiling methods that reveal it.

Software in the physical world

Aviation has fascinated me for as long as I can remember. I earned my private pilot's license in high school, and I'm drawn to aircraft, aerospace systems, and the infrastructure that connects people and places. I want to work on projects where engineering decisions have tangible, real-world effects.

Engineering that lasts

I value software that is dependable and built to last. I like to explore not only whether a system works, but why it works, where its limits are, and how to make it more robust.

Let's talk

I'm looking for engineering opportunities in aerospace, distributed systems, and performance-critical software in the greater Boston area.