tmam 1.0.0

OpenTelemetry-native Auto instrumentation library for monitoring LLM Applications and GPUs, facilitating the integration of observability into your GenAI-driven projects

🌟 tmam‑python‑sdk

tmam‑python‑sdk is an OpenTelemetry-native auto-instrumentation library for GenAI applications—providing seamless observability for Large Language Models (LLMs), GPU workloads, vector databases, and agent-based frameworks. With one line of integration, you can add powerful telemetry to your AI/ML stack.

---

🚀 Key Features

• ✅ Zero-code auto instrumentation for LLMs, GPUs, vector DBs, and agents
• 📊 Traces, metrics, and logs for LLM prompts, latency, token usage, GPU utilization, and more
• 🔌 Simple integration using a hosted TMAM collector endpoint
• 📦 OpenTelemetry-native, vendor-agnostic support for downstream observability tools
• 🔍 Supports popular frameworks like OpenAI, Hugging Face, LangChain, PyTorch, NVIDIA CUDA, and more

---

🧪 Installation

pip install tmam



Quickstart

Add observability to your GenAI application with one line:

import tmam

tmam.init(
    url="http://api.tmam.ai/api/sdk/v1",
    public_key="pk-tmam-0edeba2a-f6f3-4efd-982c-412adbb03046",
    secrect_key="sk-tmam-b320dda9-e36d-4eac-8ac5-4793fd38e002",
)

# Your LLM or agent code here
from openai import OpenAI
# ...
Once initialized, tmam will auto-instrument supported components and begin sending traces and metrics to TMAM’s backend.