M

Magellan Technology Research Institute (MTRI) Tech Stack

AI research institute building domain-specific NLP and agentic systems

Research Services Meguro-ku, Tokyo 2–10 employees Privately Held

MTRI is a Tokyo-based research institute developing specialized AI for accuracy-critical domains. The tech stack reveals a production-focused operation: RAG systems (Pinecone, FAISS, Weaviate, Milvus), model training infrastructure (PyTorch, TensorFlow, Hugging Face, Weights & Biases), and multi-cloud deployment (AWS, GCP, Azure). Active hiring across engineering and research roles in Japan and Southeast Asia, combined with documented pain points around data privacy and system reliability, suggests MTRI is moving from research prototypes toward deployable solutions in regulated verticals.

Tech Stack 32 technologies

Core StackRAG Python Vue Django FastAPI Flask AWS CloudFormation Kubernetes Docker PyTorch TensorFlow Hugging Face LangChain Pinecone Weaviate Weights & Biases GCP AWS CLI IAM Deployment Manager AutoGen OpenAI Assistants FAISS Milvus Azure Bitbucket Pipelines AWS App Runner AWS ECS AWS EKS+2 more

What Magellan Technology Research Institute (MTRI) Is Building

◆Challenges

Data privacy compliance
Compliance-driven high-stakes domains
System reliability improvement
Frontend-backend collaboration

▲Active Projects

Proof-of-concept projects
Llm fine-tuning
Ai agent system development
Rag pipeline optimization
Retrieval-augmented generation research
Vision-language models research
Ai agent systems
Rag pipelines

Hiring Activity

Accelerating15 roles · 9 in 30d

Department

Engineering

10

Research

4

Ops

1

Seniority

Mid

5

Senior

5

Intern

2

Manager

2

Junior

1

Company intelligence

Find more companies like Magellan Technology Research Institute (MTRI) by tech stack, pain points and active projects

Get started free

About Magellan Technology Research Institute (MTRI)

MTRI is a research institute focused on applied AI—specifically natural language processing, agentic systems, multimodal reasoning, and information retrieval. The organization emphasizes domain reliability and contextual accuracy over general-purpose model capability. The team is structured around engineering and research functions, with active expansion in Japan, Singapore, and Malaysia. Projects center on RAG pipeline optimization, LLM fine-tuning, and vision-language model research, indicating a shift toward operationalizing specialized AI systems rather than foundational model work.

HeadquartersMeguro-ku, Tokyo

Company Size2–10 employees

Hiring MarketsSingapore, Japan, Malaysia

Frequently Asked Questions

What is MTRI's tech stack?

MTRI uses Python, PyTorch, TensorFlow, and Hugging Face for model work; RAG frameworks (Pinecone, FAISS, Weaviate, Milvus); FastAPI and Django for backends; and multi-cloud infrastructure across AWS, GCP, and Azure for deployment and orchestration.

What are MTRI's main research areas?

MTRI focuses on RAG pipelines, LLM fine-tuning, agentic AI systems, vision-language models, and information retrieval automation—with emphasis on accuracy and reliability in domain-specific applications.