Local MLOps Foundation

A lightweight local MLOps infrastructure foundation for building, experimenting with, and monitoring machine learning and LLM applications. It includes user-friendly web UIs such as Grafana for metrics monitoring and Floci UI for S3 storage management. Run S3-compatible workflows locally without an AWS account or cloud costs.

🎯 Project Purpose & Scope

This repository provides a pre-configured local infrastructure foundation for your ML and LLM projects:

flowchart TD
    App["🚀 Your ML / LLM Project"]
    Prometheus["📊 Prometheus"]
    Grafana["📈 Grafana Dashboard"]
    S3["🪣 Local S3 Storage (Floci)"]

    App -->|Application & Model Metrics| Prometheus
    Prometheus --> Grafana
    App -->|Data, Models & Artifacts| S3

Instead of spending hours setting up local S3 storage, Grafana dashboards, and container monitoring, clone this repository and build your ML/LLM pipelines on top of it.

✨ What You Get

Component	Technology	Purpose
Local S3 Storage	Floci (`:4566`)	S3-compatible object storage for datasets, model weights, and embeddings.
S3 Management Web UI	Floci UI (`:8080`)	Visual browser for inspecting buckets and object files.
Metrics Visualization	Grafana (`:80`)	Pre-loaded dashboards for real-time container and host hardware monitoring.
Metrics Storage	Prometheus (`:9090`)	Stores and queries application, container, and host metrics.
Telemetry Collectors	Telegraf & Node Exporter	Cross-platform container and host hardware metrics collectors (fully WSL2 compatible).

🚀 Quick Start

1. Prerequisites

Docker Engine & Docker Compose v2 installed.
AWS CLI v2 (Optional, recommended for local S3 CLI testing).

Note

For step-by-step installation guides (Docker Engine WSL2, AWS CLI v2), see the Optional Documentation section.

2. Start Infrastructure

Launch all services in detached mode:

docker compose up -d

3. Verify Health Status

Check container status (use -a to view all containers, including completed init workers):

docker compose ps -a

🔍 View Expected Output

NAME                   IMAGE                       COMMAND                  SERVICE          STATUS
local_aws_floci        floci/floci:1.6.0           "/usr/local/bin/dock…"   floci            Up (healthy)
local_dashboard_init   python:3.11-alpine          "python3 -c ..."         dashboard-init   Exited (0)
local_docker_metrics   telegraf:1.34.0-alpine      "/bin/sh -c 'cat <<..."  docker-metrics   Up (healthy)
local_floci_ui         floci/floci-ui:0.2.0        "./server"               floci-ui         Up
local_grafana          grafana/grafana:13.1.3      "/run.sh"                grafana          Up (healthy)
local_nginx            nginx:1.27.4-alpine         "/docker-entrypoint.…"   nginx            Up (healthy)
local_node_exporter    prom/node-exporter:v1.9.0   "/bin/node_exporter …"   node-exporter    Up (healthy)
local_prometheus       prom/prometheus:v3.2.1      "/bin/prometheus --c…"   prometheus       Up (healthy)

🌐 Access Endpoints

Service	Access URL	Default Credentials / Notes
Grafana Dashboard	`http://localhost`	`admin` / `admin` (Pre-configured metrics dashboard)
Floci S3 Web UI	`http://localhost:8080`	S3 Management Console
Local S3 Endpoint	`http://localhost:4566`	S3 API (Key: `test` / Secret: `test` / Region: `us-east-1`)
Prometheus Metrics	`http://localhost:9090`	Direct PromQL query browser

📊 Visual Dashboards

1. Grafana Telemetry Dashboard (`http://localhost`)

Pre-provisioned dashboard displaying real-time host RAM/CPU gauges, container memory bars, CPU time-series, and network bandwidth.

🔍 Expand to view additional Grafana Panels (CPU, Network & Summary Table)

Container CPU & Network Performance

Container Summary Table

2. Floci S3 Web Console (`http://localhost:8080`)

Web interface for creating buckets, browsing folders, uploading, and downloading local S3 objects.

3. Prometheus Metrics & PromQL Browser (`http://localhost:9090`)

🔍 Expand to view Prometheus Query Browser Preview

Direct PromQL query interface for inspecting metrics, evaluating target health, and testing time-series queries.

🪣 Using Local S3 Storage & Code Integration

Interact with local S3 via AWS CLI or Python SDKs by pointing to http://localhost:4566:

1. AWS CLI Examples

# Create a bucket
aws --endpoint-url=http://localhost:4566 s3 mb s3://mlops-data

# List buckets
aws --endpoint-url=http://localhost:4566 s3 ls

# Upload datasets or model checkpoints
aws --endpoint-url=http://localhost:4566 s3 cp model.pt s3://mlops-data/v1/model.pt

# Download artifacts
aws --endpoint-url=http://localhost:4566 s3 cp s3://mlops-data/v1/model.pt ./model_downloaded.pt

2. Connect Your ML / LLM Project (Python / Boto3)

Connect your Python scripts, PyTorch models, or LangChain pipelines directly to local S3:

import boto3

s3 = boto3.client(
    "s3",
    endpoint_url="http://localhost:4566",
    aws_access_key_id="test",
    aws_secret_access_key="test",
    region_name="us-east-1"
)

# Upload model weights or dataset splits
s3.upload_file("model.pt", "mlops-data", "v1/model.pt")

🛑 Stop / Reset Stack

Stop Services (Preserve Data)

Stops containers while preserving all S3 bucket files, Prometheus metrics, and Grafana settings:

docker compose down

⚠️ Hard Reset (Delete All Stored Data)

docker compose down -v

Warning

Running this removes persistent Docker volumes and permanently deletes all stored S3 objects, datasets, model checkpoints, and Prometheus metrics.

📚 Optional Documentation

: Technical deep dive into Docker Engine API integration, Prometheus scrape flow, dashboard-init worker, and volume persistence.
: Step-by-step guide to installing standalone Docker Engine natively on Windows WSL2.
: Installing AWS CLI v2 and auto-routing commands to local Floci S3.