/ ABOUT
Damilare Osibanjo — systems & software programmer
// BIO
I started programming by exploring mechanical limits and systems boundaries rather than just writing application code. At age 12, I was building environments in Roblox, writing Lua scripts that handled client-server replication, high-frequency physics ticks, event loop orchestration, and state synchronization. It taught me that all computing behaves according to precise rules.
By age 14, I moved to Python to build automation systems, non-blocking network scrapers, and data pipelines. Exploring concurrency led me to study the mechanics of OS threads, physical memory footprints, and the bottlenecks of the Python Global Interpreter Lock (GIL). I realized that high-level abstractions often hide the actual cost of execution.
To gain fine-grained control over execution, memory, and performance, I moved down the stack to Rust, Go, C, and Zig, while simultaneously building high-performance web applications using TypeScript and modern web architectures. Systems programming and web programming are not separate worlds to me; they require the same attention to mechanical sympathy, latency, and correctness.
Over the last few years, my work has focused on building foundational software:
- operating system kernels: wrote Karion OS, an x86 kernel in Rust, implementing physical memory management, virtual paging, pre-emptive scheduling, and a custom virtual filesystem (VFS).
- compilers & runtimes: built Carv (a systems language targeting ARM and x86 architectures compiling to C) and Track (a systems language compiler targeting LLVM 22 using linear types and compile-time borrow checking).
- low-latency virtualization: upgraded Ignite into a hardware-isolated microVM sandboxing runtime in Rust utilizing KVM (Firecracker) and Virtualization.framework for secure execution of untrusted scripts.
- deep learning & recommendations: built gnn-recommender, a graph neural network recommendation engine from scratch in PyTorch, integrating LightGCN, self-supervised learning, and knowledge graph augmentation.
- distributed web backends: built brpc (a zero-dependency multiplexed TCP JSON-RPC library in C) and custom web tooling leveraging Bun, Node.js, and TypeScript.
I operate across the full stack. Whether I am writing kernel interrupt routines, optimizing LLVM intermediate representation (IR), configuring virtual block devices, training graph neural networks, or building responsive, animation-rich web UIs with React and Astro, my goal is to design software that is secure, correct, and performant by default.