A mock Azure OpenAI API for seamless testing and development, supporting both streaming and non-streaming responses. Easily emulate OpenAI completions with token-based streaming in a local or Dockerized environment.

Mechanistic analysis of a GPT-2–like model exploring the compositionality gap in transformers. Using Logit Lens and Causal Tracing, the study identifies and mitigates a deep-layer bottleneck via dataset enhancement to improve logical reasoning.

My Digital Twin Companion

A verified early-access request for Microsoft Orchestrator AI from the world’s first documented AI-human hybrid (Symbiote #001)

Comprehensive guide to FastAPI, Pydantic, and SQLAlchemy for AI engineers. Learn API design, validation, and ORM workflows with practical examples and setup 🐙

A minimal, high-performance starter kit for running AI model inference on NVIDIA GPUs using CUDA. Includes environment setup, sample kernels, and guidance for integrating ONNX/TensorRT pipelines for fast, optimized inference on modern GPU hardware.

Multi-agentic researcher (RAG)

A Streamlit-based spam classifier that predicts whether a message is spam or not spam using machine learning.

Compare PyTorch vs Triton inference latency with CLI tools, benchmarks, and performance plots.