For DevelopersSolutionsWhy Use DeepFellow?
Contact

Simplito sp. z o.o.

1-3 Grudziądzka

87-100 Toruń, Poland

KRS 0000305883

VAT EU: PL9562217643

Share Capital: 336 100 PLN

  • For Developers

  • Integration Patterns
  • Fast-track for AI Product Makers
  • Documentation
  • Industry

  • Financial Services
  • Healthcare
  • Legal
  • Government & Public Sector
  • Use Case

  • Secure AI Deployment
  • Compliance & Auditability
  • Fast-track for AI Products Makers
  • Resources

  • About us
  • Contact
  • Blog
  • Free License
  • Terms & Privacy
  • Privacy Policy

Copyright © 2025 All rights reserved. Simplito. sp. z o.o.

  1. DeepFellow
  2. for developers

For Developers

ServerInfraExtensions

DeepFellow provides all essential components for deploying production-grade AI infrastructure on your terms. Built and tested by a team of experts in the field of data privacy, with production-grade code and predefined integration workflows, designed for data-sensitive and compliance-driven environments.

DeepFellow Server

The Control Center for Your AI Infrastructure

DeepFellow Server is the main orchestration layer, responsible for all user interactions, access management, and workflow coordination. Built on FastAPI, it provides an OpenAI-compatible API that makes migration seamless.

One moment please...
One moment please...

Extensions & Plugins

Modular Security and Compliance

DeepFellow's plugin architecture allows fine-grained control over every aspect of your AI system. Add filtering, anonymization, evaluation, and logging capabilities through simple, configurable plugins.

Plugins are flexible and provide numerous integrations, such as (but not limited to):

  • Prompt Sensitive Data Detection: Control what data reaches your models through configurable filtering pipelines with automatic detection and handling of sensitive data
  • Corporate Filtering Plugin: Predefined sets of corporate data rules, enforcing organizational data policies
  • Prompt Anonymization: Protect privacy by anonymizing sensitive information before processing, using chosen tools and custom rules
  • Response Evaluation: Validate model outputs before delivery to ensure quality and safety, with industry-standard quality metrics and factual validation
  • Content Safety Validation: Verify for content safety with set of custom rules
  • Audit Logging: Complete transparency and accountability for regulatory compliance, with OpenTelemetry
  • Performance analysis: Complete AI environment analysis with Graphana and Prometeus support for load and performance verification
One moment please...
One moment please...
One moment please...
Loading...
One moment please...
One moment please...

Key Capabilities:

  • OpenAI API Compatibility: Drop-in replacement for OpenAI, Claude, and other LLM APIs
  • Access Management: Role-based access control (RBAC) for models, data, and toolboxes
  • Request Routing: Intelligent routing to appropriate infrastructure nodes
  • Workflow Orchestration: Pre-built pipelines and plugins for filtering, validation, and response evaluation
  • Load Balancing: Automatic distribution across multiple infra nodes
  • Logging & Monitoring: OpenTelemetry integration for complete observability

Installation:

curl deepfellow.ai/install.sh | bash
deepfellow server install
deepfellow server start

Deep dive into installation details

DeepFellow Infra

Where Your AI Models Run

DeepFellow Infra is the infrastructure component that hosts and serves your AI models. It supports multiple inference engines, model formats, and deployment topologies – all while maintaining security and performance.

Infrastructure Features:

  • Multi-Node Clusters: Connect multiple infrastructure nodes for scalability
  • Load Balancing: Automatic selection of available nodes with matching models
  • Model Auto-Discovery: Infra nodes automatically register available models
  • Easy setup: Infra speeds up configuration and maintains whole installation process
  • Multiple Inference Engines: Support for vLLM, Ollama, llama.cpp, and more
  • Model Catalog: Maintained catalog of tested and compatible models
  • Custom Models: Configuration-based support for your own models
  • Hybrid Deployment: Mix on-premises and cloud infrastructure
  • Widest Hardware Support: Infra supports MacOS, Linux, and Windows machines, allowing you to use all existing infrastructure to build powerful local AI systems
  • NightShift Mode: Extend capabilities of used model by Fine Tuning performed on your Infra in idle time and make your models know your domain better and better every day
df-bgs-d 3.jpg

Supported Model Providers:

  • Hugging Face: Direct integration with thousands of open-source models
  • Ollama: Easy deployment of quantized models
  • vLLM: High-throughput inference for production workloads
  • llama.cpp: Efficient CPU-based inference
  • Custom Models: PyTorch, TensorFlow, and custom frameworks

Installation

deepfellow infra install
deepfellow infra service install ollama
deepfellow infra model install ollama gemma3:1b
deepfellow infra start

Deployment Topologies

Perfect for development and testing.


  • Components: Single server + single infra node
  • Setup Time: 5 minutes
  • Use Case: Development, testing, proof-of-concept

Gaming Laptop will be enough.

Scalable production deployment.


  • Components: Server cluster + multiple infra nodes
  • Features: Load balancing, high availability, fault tolerance
  • Use Case: Production deployments, enterprise applications

Use existing infrastructure for lighter models and dedicated nodes (like NVIDIA Spark) for heavy computations

Best of both worlds – local + cloud.


  • Components: On-premises server + mixed infra (local + cloud)
  • Strategy: Sensitive data local, general tasks cloud
  • Use Case: Gradual migration, compliance + performance

Scale demanding models with cloud, protect your data on-premise.

Maximum scale and resilience.


  • Components: Multiple server instances + distributed infra across regions
  • Features: Geographic distribution, disaster recovery, global scale
  • Use Case: Large enterprises, multi-region deployments

Model Support

DeepFellow supports the latest models from Hugging Face, Ollama, vLLM, and more, including Llama, Mistral, Gemma, DeepSeek, and custom models. Deploy text, audio, and vision models quickly and securely, with hardware recommendations for every setup. Flexible architecture lets you use pre-trained, open-source, or in-house models – all with full API compatibility.

See models Documentation

Integration Ecosystem:

MCP (Model Context Protocol)

Secure Toolbox Integration


DeepFellow provides secure integration with MCP servers, enabling your AI to use corporate tools and sources while maintaining data privacy.

  • Custom Toolboxes: Create your own tool collections with well-managed access
  • Secure Connectors: Ensure compatibility without exposing internal data
  • Marketplace Ready: Compatible with growing MCP ecosystem

Vector Databases (VDB)

RAG and Knowledge Base Integration


Integrate vector databases for retrieval-augmented generation (RAG) without compromising security.

RAG Capabilities:

  • Real-time document retrieval
  • Similarity search on embeddings
  • Source tracking for factual validation
  • Secure internal knowledge base access

Data Sources & Tools

Secure Access to Internal Data


AI can operate on your organization's data through secure, server-side integrations provided by custom plugins.

  • Databases: PostgreSQL, MySQL, MongoDB, etc.
  • File Storage: S3, MinIO, NAS, local filesystems
  • APIs: Internal REST/GraphQL APIs
  • Enterprise Systems: CRM, ERP, HRIS integrations
  • Workflow Tools: n8n integration for complex workflows

Platform Features: Multi-Architecture Support

window-10-logo-window-10-icon-transparent-free-png.png
Linux-Logo-PNG-Clipart-Background.png
icons8-kubernetes.png
4844483.png
nvidia-logo.png
Apple_logo_black.svg.png
AMD_logo_pre-2013.svg.png

Feature Matrix
Core features

Infrastructure:


  • DeepFellow Server
  • DeepFellow Infra
  • Model hosting automation
  • Automatic model updates
  • Model catalog access
  • Custom model support
  • Load balancing
  • Secure server-infra communication

Security:


  • Rights management & RBAC
  • Attack-proof architecture
  • Filtering interfaces
  • Anonymization interfaces
  • Response evaluation interfaces
  • Content safety check interfaces
  • Log system interface
  • EU AI Act compliance

Integrations:


  • OpenAI API compatibility
  • MCP server support
  • Vector database integration
  • Hugging Face models
  • Multiple inference engines

Enterprise Features
& Add-Ons


  • Infra Manager (unified management portal)
  • Kubernetes auto-deployment
  • Custom model catalog server
  • Corporate filtering plugins
  • Corporate credentials (OpenLDAP, etc.)
  • Confidential computing integrations
  • Remote computing integrations
  • Priority support & SLA
  • Custom plugin development
  • Dedicated training & onboarding
  • Architecture consulting

Ready to deploy trustworthy AI?

Book a demo of DeepFellow – we will show you how it works in action.

Explore resources about licensing, use cases, and technical documentation.

Schedule a demoSee documentationExplore use casesRead the license