DEV Community

Taki
Taki

Posted on

Open source AI stack components

Here’s a comprehensive and categorized list of open source AI stack components that you can mix and match when building GenAI applications β€” especially when focusing on modularity, scalability, and performance. This includes components for data processing, model serving, retrieval-augmented generation (RAG), vector search, and orchestration.


🧠 Foundational Model Alternatives

Models you can self-host or fine-tune:

  • LLMs

  • Multimodal

    • llava – Language + vision.
    • bakllava – More optimized multimodal variant.
    • CLIP – Text-image understanding.
  • Fine-Tuning

    • QLoRA, LoRA, PEFT (via πŸ€— Transformers + PEFT)
    • Axolotl – Full stack fine-tuning.

πŸ“š RAG (Retrieval-Augmented Generation) Stack

Tools to power knowledge-based Q&A systems:


πŸ”§ Serving & Orchestration

Serving models with APIs, managing prompts, memory, and chaining tools:


πŸ–ΌοΈ Frontend / Chat UI

For chatbots or multimodal interfaces:

  • Next.js – UI + SSR/ISR.
  • ShadCN/ui – Design system for building clean UIs.
  • Chatbot UI – Open-source ChatGPT-style interface.
  • Open WebUI – Web UI for LM Studio / Ollama.

πŸš€ Inference & Runtime Optimization

  • llm.rs – LLM inference in Rust.
  • ggml – Quantized models, runs on CPU.
  • exllama – High-perf quantized inference.

πŸ”’ Security & DevOps (for production)

  • AuthN/AuthZ: [Auth.js (NextAuth)], [Clerk], [Ory], [ZITADEL]
  • Logging/Tracing: [Langfuse], [OpenTelemetry], [Sentry]
  • DevOps: Docker, Kubernetes, GitHub Actions, Terraform

🧱 Full Stack Boilerplates

If you're looking to start fast:


πŸ§ͺ Experimental Tools

  • Ollama – Run and manage LLMs locally.
  • Modal – Serverless infra for AI.
  • LiteLLM – Drop-in proxy for OpenAI-compatible APIs.

Top comments (1)

Collapse
Β 
dukeastro93 profile image
Duke Astro β€’

Could you write a new topic about how to use that