Mandla Mkhatshwa@mandlalucky
3
Events attended
1
Submissions made
South Africa
3+ years of experience
About me
I am Mandla Lucky Mkhatshwa a full-stack developer who is a final-year Computer Science student ,closing in on my diploma after completing my Matric in 2021. My track record shows a heavy focus on enterprise and mobile architecture. I have actively built and deployed systems like the Witbank QueueMaster using React and Spring Boot, engineered PostgreSQL database architectures for a Real Estate Management System, and mapped out the Jetpack Compose structure for the Frontline Intelli-care mobile app. I frequently collaborate on these large-scale system deployments alongside my team. Beyond the IDE, I am mapping out a rigorous study strategy for my AWS Developer Associate and Solutions Architect Professional exams. When I step away from the code, I am dropping into Call of Duty: Warzone, actively fine-tuning my mouse DPI between 800, 1200, and 1600 to optimize my target tracking. In short: I am a full-stack engineer, a precision gamer, and a Gemini currently pushing through the final stretch of an AMD hackathon.
🤓 Latest Submissions
GemmaForge
The high-performance computing industry faces a massive crisis: monopolistic hardware dependency. Critical infrastructure in FinTech, AI, and enterprise computing is trapped inside NVIDIA’s closed-source CUDA ecosystem due to legacy codebase dependencies. Migrating to cost-effective AMD hardware is a massive bottleneck. Manual rewrites are slow, expensive, and highly prone to "silent architecture failures"—such as incorrectly mapping NVIDIA's 32-thread warps to AMD's 64-thread physical wavefronts. GemmaForge acts as an intelligent, automated migration bridge. It goes far beyond simple syntax swapping. Powered by a native vLLM container running the 12B parameter Gemma model directly on AMD Instinct hardware, GemmaForge actually understands structural GPU architecture. It automatically recalculates underlying thread and lane mathematics, handling hardware-specific divergence patterns to output fully compliant AMD ROCm HIP C++ code. To make this safe for highly regulated corporate environments, GemmaForge wraps this advanced AI translation engine inside a secure, production-ready DevOps dashboard built with Spring Boot and Angular. It features strict Role-Based Access Control (RBAC), separating "Engineers" (who trigger migrations and run debug tests) from "Reviewers" (who audit AI security logs and approve code states). Every action, token count, and codebase delta is permanently tracked in a PostgreSQL ledger. GemmaForge shifts hardware modernization from a risky software engineering burden into a fast, transparent, and profitable capital upgrade.
13 Jul 2026
