Gaurang Vishwakarma
Specializing in ML Systems, Distributed Systems, and Infrastructure
Production C++ & Python • Distributed LLM Training • Kubernetes & HPC • Open Source Systems at Scale
Technical Skills
AI and ML
- PyTorch
- LLMs
- LoRA
- Fine-Tuning
- RAG
- GPU Clusters
HPC and System Tools
- MPI
- Slurm
- CUDA/ROCm
- Linux
- Kubernetes
- Ansible
Development
- Modern C++
- Python
- React/Next + JavaScript
- Bash Shell Scripting
- CI/CD
- Docker/Singularity
Systems Engineering
Building ML Systems at Scale
Focused on building and optimizing large-scale ML systems, distributed computing infrastructure, and production-grade tooling for high-performance workloads. Experience spanning modern ML frameworks, container orchestration, and HPC clusters.
Core Expertise
- ML Systems: Distributed LLM training, LoRA fine-tuning, RAG architectures, PyTorch optimization
- Infrastructure: Kubernetes, Docker, Slurm, Terraform, Ansible, CI/CD pipelines
- Systems Programming: Production C++ and Python, Bash automation, performance optimization
- Open Source Leadership: Maintainer of Linux distribution serving 500K+ users globally
HPC & Accelerator Infrastructure
Production experience across multiple world-class supercomputing clusters: ARCHER2, Cirrus, DKRZ Levante, PSC Bridges-2, EIDF GPU Cluster, EIDF Cerebras Cluster.
Hands-on benchmarking and optimization across diverse hardware accelerators including NVIDIA A100, H100, H200, AMD MI210, MI300X, and Cerebras CS-3.
Featured Projects
LLM LoRA Fine-Tuning & Performance Optimization
Distributed LoRA fine-tuning pipelines for large language models, optimized for performance across heterogeneous accelerator clusters. Focused on scalable training, reproducibility, and benchmarking across GPU and wafer-scale systems.
Arka Linux GUI - Open Source OS
Open-source Linux operating system and GUI stack used by 500,000+ users globally as a daily driver. Focused on system reliability, modular build pipelines, and long-term maintainability across diverse hardware. Features a vibrant community on our support platforms.
Scholar Sense - RAG System
Production-grade Retrieval-Augmented Generation (RAG) system for large-scale document analysis. Built scalable NLP pipelines for embedding, indexing, and LLM-based inference using containerized microservices.
oschat - Real-Time Communication Platform
High-performance real-time communication platform supporting thousands of concurrent WebSocket connections. Designed for low-latency messaging, horizontal scalability, and production cloud deployment.
Beyond Engineering
Outside of building distributed systems and optimizing ML pipelines, I'm a performing drummer with 30+ live shows across indoor venues and arenas. Music brings the same creative problem-solving and rhythm that drives great engineering.
