RITS AI Stack Overview

Last updated: 2025-07-11

1 · Quick Facts

Metric	Value
Total GPUs	9
Machines	3 (Workstation × 1, Alienware × 2)
GPU Models Represented	7 (5090, 4090, 3090, 4070, 3060 Ti, 2080 Ti, 2060 Super)
Core Services Hosted	Ask System (LLM/TTS/STT/API), Create Image, Whisper STT, Chat Demo
AI Models	Qwen3-30B-A3B-AWQ, Sana, Kokoro-82M, whisper-large-v3

2 · Hardware Detail

Machine	Location / Form‑factor	GPU Configuration	Notes
Workstation	Internal × 2 / eGPU × 3	5090 × 1 (internal) 4090 × 1 (internal) 4090 × 1 (eGPU) 3090 × 2 (eGPU)	5‑GPU flagship compute node
Alienware nyush2920wk	Tower	4070 × 1 3060 Ti × 1	Mid‑range support node
Alienware nyush2290wk	Tower	2080 Ti × 1 2060 Super × 1	Legacy support node

3 · Software & Service Map

Service Group	Public Endpoint(s)	Primary Components	Host Machine
Ask System	https://ask-angie.ritsdev.top https://ask-roary.ritsdev.top	LLM, TTS, STT API gateway	Workstation (LLM/TTS/STT) IT server (API gateway)
Create System	https://create-image.ritsdev.top	Image Generation API	Alienware nyush2920wk
Whisper STT	https://whisper.ritsdev.top	Speech‑to‑Text (Whisper)	Alienware nyush2920wk
Chat Demo	https://chat.workstation.ritsdev.top	Front‑end demo for Ask System	Workstation

4 · Topology Snapshot

        flowchart TD
    subgraph Workstation
        A5090["5090 (internal)"]
        A4090_int["4090 (internal)"]
        A4090_egpu["4090 (eGPU)"]
        A3090a["3090-1 (eGPU)"]
        A3090b["3090-2 (eGPU)"]
    end

    subgraph nyush2920wk
        B4070["4070"]
        B3060["3060 Ti"]
    end

    subgraph nyush2290wk
        C2080["2080 Ti"]
        C2060["2060 Super"]
    end

    AskLLM["Ask LLM / TTS / STT"] -- uses --> Workstation
    AskAPI["Ask API"] -- connects --> AskLLM
    CreateImage["Create Image API"] -- runs on --> nyush2920wk
    Whisper["Whisper STT"] -- runs on --> nyush2920wk
    ChatDemo["Chat Demo"] -- front-end --> Workstation
    

5 · AI Models & Specifications

Service	Model	Source	Specifications	Task
LLM	Qwen3-30B-A3B-AWQ	https://www.modelscope.cn/models/swift/Qwen3-30B-A3B-AWQ	30B parameter MOE model with AWQ quantization for efficient inference	language processing
STT	Whisper Large V3	https://huggingface.co/openai/whisper-large-v3	OpenAI’s largest Whisper model for high-accuracy speech recognition	speech recognition
TTS	Kokoro-82M	https://huggingface.co/hexgrad/Kokoro-82M	82M parameter text-to-speech model optimized for natural voice synthesis	speech synthesis
Image Generation	Sana	https://huggingface.co/collections/Efficient-Large-Model/sana-673efba2a57ed99843f11f9e	Efficient large model for high-quality.	image generation

---

Feedback