I'm Elclark Kuhu, a software engineer focused on high-performance web applications and system architectures. From building automated binary asset extraction and transcoding pipelines to writing hardware telemetry bridges, running sub-millisecond machine learning models on Node.js, and crafting fast SvelteKit web interfaces.
Software Engineer with a Computer Science background focused on high-performance web applications and system-level architectures.
I specialize in designing effective, low-friction systems, spanning from streamlined full-stack developer workflows to connecting serverless edge ecosystems directly to native hardware interfaces via protocols like WebSockets, FFI, and Win32.
My experience includes architecting custom LLM repository context tools, real-time hardware telemetry pipelines in TypeScript & SvelteKit, and training/deploying edge machine learning models with Python, XGBoost, and ONNX Runtime.
Integrating C++/C# via FFI interop, binary asset conversion (Protobuf), and sub-millisecond edge API routing.
Deploying ML models with ONNX Runtime and XGBoost for real-time sensor forecasting and automated alerts.
Event-driven bridges, Win32 API, DDC/CI monitor control, and atmospheric sensor arrays.
Automated ingestion and reverse-engineering pipeline designed to decrypt and process layered proprietary game clients. Features launcher protocol emulation, multi-tiered cryptographic unpacking of binary catalogs and encrypted SQLite databases, high-throughput C++ FFI transcoding (lossless WebP for 2D skeletal rigs, direct FSB-to-Ogg audio bitstream remuxing), and a SvelteKit analytical research engine deployed on Cloudflare Workers.
Custom IoT sensor array and smart lighting automation platform. Features an event-driven telemetry bridge over WebSockets that synchronizes ambient light and atmospheric data in real time, a native Windows client (C#/.NET 10) that dynamically controls physical monitor brightness via DDC/CI based on live room lux levels, and an XGBoost rainfall prediction pipeline deployed with ONNX Runtime for sub-millisecond local weather inference.
High-performance C++ CLI utility that bundles entire multi-file repositories into optimized Markdown structures for LLM ingestion. Features an automated SQLite schema extraction tool that generates relational definitions to accelerate LLM contextual comprehension.
Open for full-time software engineering roles, technical contract work, and system architecture collaborations. Located in Manado, Indonesia.