Aakarsh Kashyap

About

I am Aakarsh Kashyap, an inference engineer. I publish under the handle souls-syntax. I am reading for the degree of B.Tech in Computer Science and Engineering at GLA University, concurrently with a B.S. in Data Science and Artificial Intelligence from IIT Madras.

My practice is inference engineering: the work of making trained models execute efficiently and predictably on constrained hardware. That practice is built on low-level systems engineering — GPU compute, compiler infrastructure, operating systems, and runtime development. I work principally in C++, x86-64 assembly, and CUDA, with production experience in Go and Python across backend and distributed systems.

I maintain soft-cuda, a hybrid CPU/GPU ahead-of-time deep-learning framework with a three-pass compiler, a bump allocator, and a lazy evaluation directed acyclic graph. The hybrid dispatch model yields a 35-percent throughput improvement over the GPU-only baseline, measured at 0.94 ms per step against 1.44 ms per step for the reference configuration. The full project write-up and five others are catalogued under projects.

I intend to practise in GPU compute or graphics engineering, as professional work.

Skills

• languages: C / C++, Go, Python, x86-64 ASM, CUDA, SQL
• systems: OS internals, memory allocators, POSIX APIs, IPC, syscall interface
• GPU / HPC: CUDA kernel programming, CUDA streams, cuBLAS, cuDNN, SGEMM, double buffering
• backend: REST, scatter-gather, write-through cache, JWT, Argon2id, RBAC, webhook verification
• tooling: Git, Docker, CMake, Make, GDB, Neovim, Arch Linux

Correspondence

Available for internship and contract work in inference engineering, systems programming, and GPU compute. The résumé and experience record is maintained in full, and notes on the work appear in the engineering journal. souls.syntax@gmail.com