Vraj Patel
@vrajpatel105
Joined on 31 May 2024
-_-
GitHub Stats
13
Followers
55
Repositories
0
Organizations
0
Gists
6
Pull Requests
0
Issues
666
Commits
0
Sponsors
1
Contributed To
0
Star Earned
Most Used Languages
81.68%
Jupyter Notebook
8.80%
Python
5.19%
HTML
1.73%
C
1.42%
C++
1.18%
Cuda
0.01%
CMake
Popular Projects
mlops-inference-service
Production-style ML model serving pipeline with FastAPI, Docker, and automated CI/CD deployment which also includes request logging, latency monitoring, and health checks
Unknown
0
0
0
0
modern-llm-from-scratch
A decoder-only transformer built from scratch with the architecture components used in current frontier LLMs such as : RMSNorm, SwiGLU, RoPE, grouped-query attention, and Mixture-of-Experts. Implemented and validated incrementally, extending the original from-scratch Transformer project into a modern architecture.
Python
0
0
0
0
paged-inference-engine
A small inference server built from scratch: paged KV cache, continuous batching scheduler, and custom FlashAttention-2 + INT8 kernels, benchmarked against a naive baseline. Design inspired by vLLM's PagedAttention (build from scratch, not derived from their codebase)
Python
0
0
0
0
cpp-gpu-inference
C++ systems programming to GPU-accelerated ML inference - CUDA, quantization, and multi-platform deployment.
Python
0
0
0
0
Generative-Models
Covering AutoEncoders, VAE, GANs from scratch
Python
0
0
0
0
Deep-Learning-Representations
Transfer Learning for nlp
Python
0
0
0
0
Top Contributions
Top contributions made by the user in the last year.
Charts
Follow Up
Activity Graph
Contributions Calendar
Contributions made by the user in the last 365 days.
Recent Activity
9/2/2026, 7:42:50 AM
9/1/2026, 7:42:23 AM
9/1/2026, 7:41:43 AM
8/31/2026, 8:57:27 AM
8/23/2026, 10:13:09 PM
8/23/2026, 9:58:50 PM
8/23/2026, 9:53:31 PM
8/23/2026, 3:57:19 AM
8/23/2026, 3:52:03 AM
8/23/2026, 3:39:10 AM
8/23/2026, 3:38:41 AM
8/23/2026, 2:57:12 AM
8/22/2026, 9:48:50 PM
8/22/2026, 12:46:09 AM