avanindra singh — Co-Founder

I am a Principal Architect and Researcher sitting at the intersection of High-Performance Computing (HPC) and Generative AI. I specialize in engineering the large-scale systems required to train, fine-tune, and serve frontier models in enterprise environments. Currently at Shell, I lead the domain-adaptation strategy for Large Language Models. I architected the internal GenAI ecosystem from scratch, scaling Agentic RAG systems to reason over millions of technical reports and orchestrating distributed training for Llama 3 and DeepSeek on 100+ GPU clusters. Beyond LLMs, I have deep roots in Scientific Computing and 3D Computer Vision. I have spent years optimizing "impossible" workloads—from achieving a 100x I/O speedup in seismic deep learning pipelines to writing custom CUDA kernels that reduced 3D Digital Rock simulation times from weeks to minutes. Core Technical Competencies: AI Infrastructure: Distributed Training (Megatron-LM, FSDP), Multi-Node Scaling, Agentic RAG. HPC & Hardware: NVIDIA DGX/HGX, InfiniBand, Slurm, CUDA Kernel Optimization, C++. Scientific HPC and AI: 4D/5D convolution and Tensor Optimization, Volumetric Rendering, Photogrammetry, Seismic Deep Learning. My passion lies in solving the "last mile" of performance—optimizing the hardware-software stack to turn experimental research into production-grade capability.

Stackforce AI infers this person is a High-Performance Computing and AI expert specializing in enterprise-level solutions.

Location: Gorakhpur, Uttar Pradesh, India

Experience: 18 yrs 8 mos