LLMs for the next
generation of intelligent applications
NGJOO AI focuses on core large language model technology, delivering a full-stack AI offering across training, content generation and security evaluation.
Explore productsCore Services
End-to-end capabilities across the large-model lifecycle
LLM Training
In-house distributed training stack supporting pre-training, fine-tuning and RL alignment for hundred-billion-parameter models across NLP and multimodal workloads.
AIGC Application Development
Content generation, AI Agent orchestration and RAG knowledge bases — helping teams ship LLM-driven business systems end to end.
AI Red Teaming & Defense
LLM red-team testing, prompt-injection defense, adversarial-sample detection and robustness evaluation — built for AI systems running in production.
Data & Evaluation
High-quality training-data engineering, model benchmarking and domain capability comparisons — data-driven inputs for every iteration cycle.
Product Portfolio
Tools and platforms built on top of large language models
Paper Collector
Automatically collects trending papers from HuggingFace, then uses an LLM to generate summaries, categories and tags — helping researchers stay on top of the latest AI work.
PDF Reader
Upload a PDF and the model extracts key points into a mind map — built for close reading of papers and fast review of technical documents.
AI-OA
17 specialist AI agents coordinate across HR, finance, IT operations and approvals, completing internal workflows through natural-language interaction.
IntelligentRouting
A transparent proxy in front of your LLM API calls: dedupes redundant context, trims RAG retrieval and compresses conversation history. One-line integration, typically cutting API spend by 30–60%.
OpenContext
A hierarchical knowledge base built for AI agents. Documents are decomposed into L0/L1/L2 summaries; agents browse them like a filesystem and fetch only what they need. Ships with session memory and retrieval traces.
VideoLearn
Import from Bilibili or upload locally. Automatic speech transcription, timestamp-aware contextual Q&A, and structured notes generated from the dialogue — so video learning leaves something durable behind.
HaoHao
A private space to talk things through. Seven specialist skill agents work together with a silent user model and semantic memory — like an old friend who listens, helps you think clearly and offers advice when asked.
AIContracts
An AI legal department for small and mid-sized businesses. RAG + LLM for contract review, risk scoring, plain-language explanation, drafting, translation and legal Q&A — every output cites the underlying clause.
CodeGuard
Security scoring for code generated by AI coding tools, grounded in 132 real CVEs with Docker-based PoC verification. Hooks into GitHub via webhook for PR-level CVE regression detection, with zero false-positive alerts.
AI Analyst
19 AI analysts — including 13 investor-personas — run five analytical dimensions in parallel and aggregate the output via LangGraph into a consolidated report. Supports real A-share and US equities data.
About NGJOO AI
Technology at the core, applied to real-world scenarios
Engineering-led, scenario-driven
NGJOO AI is an artificial intelligence company focused on the core technologies behind large language models. Our work spans LLM training and fine-tuning, AIGC application development, and AI security and red-teaming.
The team comes from AI research and production engineering. We know the full pipeline — from data processing through distributed training to deployment — and turn that into enterprise-grade solutions.
Good AI products need strong models, solid engineering and rigorous safety work. We believe all three matter.
Competition Awards
Team "Refuse to be a Chive" representing NGJOO AI at AI competitions
Contact
We would like to talk with you about AI collaboration