Chapel Parallel Computing Engineer for HPC Workloads

I write high-performance code in Chapel, the parallel language designed for productive scientific computing at scale. If your simulation, solver, or data pipeline is choking on a single core, I move it onto every core you own and then across a whole cluster. My work centres on turning ordinary numerical code into something that keeps thousands of threads busy without you rewriting everything by hand. I use Chapel's data-parallel forall loops, distributed domains, and locale model to spread arrays across nodes, and I lean on task parallelism with begin, cobegin, and sync variables where the algorithm needs finer control. Typical jobs I take on: porting a serial C or Fortran kernel to Chapel, parallelising stencil and matrix operations, building distributed hash tables and graph traversals, and profiling why a run stalls on communication rather than arithmetic. I measure everything. Before I call a job done I show you strong and weak scaling curves, so you can see how the code behaves as cores and problem size grow, and I explain where the limits come from. Deliverables include clean, commented source, a build recipe for your compiler and launcher, a short report on scaling and memory behaviour, and guidance on running it on your own hardware or scheduler. I have shipped Chapel code for Monte Carlo sampling, N-body dynamics, computational fluid steps, and large sparse linear algebra. I care about correctness first, then speed: I add reproducible test cases and compare results against a trusted reference before tuning. Tell me your node count, interconnect, and the kernel that hurts, and I will give you an honest estimate of the speedup you can expect and the effort to get there.