FedBloom — No-Code Federated Learning

Why FedBloom?

You shouldn't need a PhD
to protect your data

Federated learning was locked behind thousands of lines of Python. FedBloom opens it to clinicians, bankers, researchers, and anyone with data worth protecting.

client.py — Traditional Federated Learning THE OLD WAY

# Step 1: Learn Python (months)
# Step 2: Learn PyTorch / TensorFlow (months)
# Step 3: Learn Flower framework (weeks)
# Step 4: Write all of this...

import flwr as fl
import torch, torch.nn as nn
from torch.utils.data import DataLoader
from collections import OrderedDict

class Net(nn.Module):
def __init__(self):
super().__init__()
self.fc1 = nn.Linear(784, 128)
self.fc2 = nn.Linear(128, 10)
def forward(self, x):
return self.fc2(torch.relu(self.fc1(x)))

class FLClient(fl.client.NumPyClient):
def get_parameters(self, config):
return [v.cpu().numpy()
for _, v in
self.model.state_dict().items()]
def fit(self, parameters, config):
# ... load weights, train loop,
# ... DataLoader, optimizer,
# ... loss function, backprop ...
pass
def evaluate(self, parameters, config):
# ... more boilerplate ...
pass

# Server setup
strategy = fl.server.strategy.FedAvg(
fraction_fit=0.3,
min_fit_clients=3,
min_available_clients=3,
on_fit_config_fn=fit_config,
)

fl.server.start_server(
server_address="0.0.0.0:8080",
config=fl.server.ServerConfig(
num_rounds=10
),
strategy=strategy,
)

# Also need: data pipeline, preprocessing,
# environment setup, dependency management,
# network config, firewall rules...
# Estimated setup time: days to weeks.

Requirements

Python expertise PyTorch / TF knowledge Flower / TFF framework Network & firewall config Days to weeks of setup

FedBloom — No-Code FL THE NEW WAY

Open FedBloom → Enter as Server

No installation scripts. Just launch the app.

Click "+ New Experiment"

Name your experiment and choose a mode.

Select Model, Dataset, Strategy

Dropdowns — no code, no config files.

Upload your CSV dataset

FedBloom auto-preprocesses. No pandas. No sklearn.

Click ▶ Start

Watch accuracy, loss, and topology update live.

✓

Done. Model trained.

Export results, download model, share insights.

Who can use this

🏥 Clinicians 🏦 Bankers 🧬 Researchers 🛡 Insurers 📊 Analysts

Your data is valuable. You shouldn't need to hire a developer to use it.

FedBloom gives hospitals, banks, labs, and clinics the same federated AI tools that only well-funded tech teams had before — with no code required.

⬇ Download Free — Start in Minutes

Features

Everything you need
for federated learning

From model selection to real network deployment — all through a clean, modern GUI.

🧠

30+ Built-in Models

MLP, CNN, ResNet-18 and 9 more torchvision architectures (MobileNet, EfficientNet, VGG, DenseNet, ViT…), Logistic Regression, Linear SVM, Decision Trees, Random Forest, XGBoost, LightGBM, CatBoost, Naive Bayes, TabNet, Wide & Deep, and more — ready to use with zero config.

🌐

Real Network Deployment

Deploy server and client roles across real machines using your local network, Tailscale VPN, or public IP + port forwarding. Training continues even if not every expected client connects, and the server is notified automatically — with the option to keep going — if a client drops mid-session.

📊

Live Metrics Dashboard

Track accuracy, loss, F1 score, and per-round progress in real time with interactive charts and metric history.

🔒

Privacy-Preserving

Differential privacy (DP-SGD) and robust aggregation (Krum, FedProx) keep your federated training secure and noise-resistant.

🗂️

Data Pre-processing

A full workbench for both server and client roles — explore your data (overview, stats, charts), handle missing values, drop duplicates/outliers, normalise, one-hot/label encode, and export before training. Choose from 17+ built-in datasets (MNIST, CIFAR-10, Adult Income, Iris, Wine, and more) or bring your own — upload any CSV or image dataset.

⚡

Hyperparameter Optimisation

Automated HPO tunes learning rate, batch size, and model depth — find optimal configs without manual grid searches.

💬

Server–Client Messaging

Integrated chat between server operator and connected clients — coordinate training sessions without external tools.

⊞

Role-Aware Dashboard

Both Server and Client modes open to a tailored home screen — quick actions, recent experiments or deployments, and a live feature overview, scoped to what that role actually needs.

🎨

Dark & Light Themes

Beautiful, eye-friendly interface with instant theme switching. Looks great on any monitor, any time of day.

How it works

From launch to training
in minutes

FedBloom removes every technical barrier between you and your first federated experiment.

Choose your role

Launch FedBloom and select Server or Client. The server orchestrates the FL plan; clients contribute local data. Switch roles anytime from the welcome screen.

Configure your experiment

Pick a model, dataset, partitioner, aggregation strategy, and number of rounds — all via drop-downs and sliders. No YAML, no JSON, no terminal.

Connect clients

Share the server address. Clients enter it in the Client workspace and connect instantly — over the same WiFi, a VPN (Tailscale), or the open internet.

Train & monitor

Hit Run. Watch accuracy climb, loss fall, and client topology update live on the dashboard — then export results or share them with your team.

Pricing

Plans for every team

Start free. Upgrade when you need more clients, more models, or real network deployment.

⊞

Free

Explore federated learning at no cost

€0 /mo

Free forever · no card needed

Up to 5 simulated clients
10 core models — MLP, Logistic Regression, Decision Tree, Random Forest, XGBoost, Extra Trees, AdaBoost, Naive Bayes, Ridge, Elastic Net
Accuracy, Loss, F1, ROC metrics — global & local charts
All 9 aggregation strategies
IID, Dirichlet, Label Shard & Custom partitioners
Results & Artefacts export (CSV, ZIP, .pkl)
Local FL simulation

◈

Pro

Scale experiments and deploy to real networks

€3 /mo

Billed monthly · cancel anytime

Everything in Free, plus:
Up to 50 simulated clients
20+ extended models — CNN, DNN, ResNet-18, Linear SVM, Custom NN Builder
9 TorchVision models — MobileNetV2/V3, EfficientNet-B0, VGG-16, DenseNet-121, ViT-B/16 & more
TabNet · Wide & Deep tabular deep learning
HistGradientBoosting · LightGBM · CatBoost · LinearSVC
Differential privacy (DP-SGD via Opacus)
Hyperparameter optimisation (HPO)
Real network deployment — LAN, Tailscale, port forwarding + TLS
FL Plans — server pushes model & config to clients
Server–client in-app chat

✦

Advanced

Maximum clients, unlearning & explainability

€5 /mo

Billed monthly · cancel anytime

Everything in Pro, plus:
Up to 100 simulated clients
Federated Machine Unlearning — remove a client's data influence post-training
Explainability — XAI / SHAP feature importance
AI Assistant — local LLM via Ollama
Priority support

System Requirements

What you need to run FedBloom

Lightweight and self-contained — no cloud dependency, no Docker, no setup scripts.

🖥 Operating System

Windows 10 (64-bit) or later
Windows 11 recommended
macOS / Linux — coming soon

⚙ Hardware

CPU: 4-core x86-64 (Intel / AMD)
RAM: 4 GB minimum, 8 GB recommended
Storage: 500 MB free disk space
GPU: optional (CUDA 11+ for acceleration)

🐍 Runtime

Python 3.10+ (bundled in installer)
PySide6 / Qt 6.x (included)
No manual Python install required

🌐 Network (real deployment)

Local: same WiFi / LAN network
Remote: Tailscale VPN or port forwarding
Ports: configurable (default 8080)
Simulation: no network needed

📦 Quick install

        # Install via pip (dev mode)

        pip install -e .

        # Or run directly

        python -m fedkit

Federated Learning
Without the Code

You shouldn't need a PhD
to protect your data

Everything you need
for federated learning

From launch to training
in minutes

Plans for every team

What you need to run FedBloom

Built with passion for
privacy-preserving AI

Download FedBloom

Federated Learning Without the Code

You shouldn't need a PhDto protect your data

Everything you needfor federated learning

From launch to trainingin minutes

Plans for every team

What you need to run FedBloom

Built with passion forprivacy-preserving AI

Download FedBloom

Federated Learning
Without the Code

You shouldn't need a PhD
to protect your data

Everything you need
for federated learning

From launch to training
in minutes

Built with passion for
privacy-preserving AI