# Kejian Tong

> software engineer and AI researcher

**Wikidata**: [Q137643975](https://www.wikidata.org/wiki/Q137643975)  
**Source**: https://4ort.xyz/entity/kejian-tong

## Summary
Kejian Tong is a software engineer and artificial intelligence researcher whose work spans machine learning, deep learning, and natural language processing, including large language models. He is associated with Northeastern University and has published research on retrieval-augmented generation, question answering, and applied machine learning systems.

## Biography
- Education: Master of Science, Northeastern University
- Known for: Research publications in AI/ML and NLP, including document-level question answering and retrieval-augmented generation
- Employer(s): Northeastern University (education affiliation)
- Field(s): artificial intelligence; machine learning; natural language processing; deep learning; software engineering; large language model

## Contributions
Kejian Tong has contributed to applied AI and machine learning research through a set of publications focused on question answering, retrieval-augmented generation, and ensemble learning. His notable works include **“Enhancing Document-Level Question Answering via Multi-Hop Retrieval-Augmented Generation with LLaMA 3”** and **“An Intelligent-Aware Transformer with Domain Adaptation and Contextual Reasoning for Question Answering,”** both centered on improving question answering systems with modern transformer-based methods and adaptation strategies. He has also published on safety- and memory-oriented approaches for domain-adaptive QA, including **“Memory-Augmented Knowledge Fusion with Safety-Aware Decoding for Domain-Adaptive Question Answering.”**  
Beyond NLP, Tong has authored applied ML research in financial and decision systems, such as **“FinStack-Net: Hierarchical Feature Crossing and Stacked Ensemble Learning for Financial Fraud Detection”** and **“An Integrated Machine Learning and Deep Learning Framework for Credit Card Approval Prediction.”** Additional work extends to systems-oriented ML topics, including cloud telemetry prediction (**“Task-Gated Attentive Multi-Task Ensemble Learning for Joint Energy and Latency Prediction in Cloud Telemetry”**) and microservices root cause analysis (**“GraphRCA-Chorus: Choreographed Multi-Agent Graph Transformers for Root Cause Analysis in Microservices”**).

## FAQs
### Q: Who is Kejian Tong?
A: Kejian Tong is a software engineer and artificial intelligence researcher. His work includes machine learning and natural language processing research, with publications on question answering and retrieval-augmented generation.

### Q: What is Kejian Tong known for?
A: He is known for research papers on document-level question answering, retrieval-augmented generation (including work referencing LLaMA 3), and applied machine learning methods such as ensemble learning for fraud detection.

### Q: Where did Kejian Tong study?
A: He studied at Northeastern University and holds a Master of Science degree.

### Q: What topics does Kejian Tong work on?
A: His listed fields and interests include artificial intelligence, machine learning, natural language processing, deep learning, large language models, and software engineering, with additional interests in cloud computing, distributed systems, and microservices.

## Why They Matter
Kejian Tong’s significance comes from bridging modern NLP/LLM research with practical, systems- and application-driven machine learning problems. His publications address core challenges in question answering—such as document-level reasoning, domain adaptation, and retrieval-augmented generation—areas that are central to making large language model systems more useful for real-world information access. At the same time, his work extends beyond NLP into applied predictive modeling and ensemble learning for financial decision-making and fraud detection, demonstrating how ML techniques can be structured and combined for high-stakes domains.  
Tong’s research portfolio also reflects an interest in operational and infrastructure contexts (cloud telemetry and microservices), where ML methods are used to predict performance characteristics or support root cause analysis. Without contributions of this type—spanning QA/LLM methods and applied ML in finance and systems—practitioners would have fewer published reference points for combining retrieval, adaptation, and ensemble strategies across different problem settings.

## Notable For
- Author of “Enhancing Document-Level Question Answering via Multi-Hop Retrieval-Augmented Generation with LLaMA 3.”
- Author of “FinStack-Net: Hierarchical Feature Crossing and Stacked Ensemble Learning for Financial Fraud Detection.”
- Author of “An Integrated Machine Learning and Deep Learning Framework for Credit Card Approval Prediction.”
- Author of “GraphRCA-Chorus: Choreographed Multi-Agent Graph Transformers for Root Cause Analysis in Microservices.”
- Associated author identifiers across major scholarly indexes (arXiv author ID: **tong_k_1**; Google Scholar author ID: **JUGvC_oAAAAJ**; Semantic Scholar author ID: **2368457510**).

## Body
### Identity and Roles
- Occupations:
  - Software engineer
  - Artificial intelligence researcher (listed as preferred occupation)
- Described as: “software engineer and AI researcher.”

### Education
- Northeastern University
- Degree: Master of Science

### Fields and Interests
- Field(s) of work:
  - Artificial intelligence
  - Machine learning
  - Natural language processing
  - Deep learning
  - Large language model
  - Software engineering
- Interested in:
  - Artificial intelligence; machine learning; natural language processing; deep learning; large language model
  - Distributed system; cloud computing; microservices

### Selected Publications (Notable Works)
- Question answering, RAG, and transformers:
  - “Enhancing Document-Level Question Answering via Multi-Hop Retrieval-Augmented Generation with LLaMA 3”
  - “An Intelligent-Aware Transformer with Domain Adaptation and Contextual Reasoning for Question Answering”
  - “Memory-Augmented Knowledge Fusion with Safety-Aware Decoding for Domain-Adaptive Question Answering”
  - “Tool-Augmented Hybrid Ensemble Reasoning with Distillation for Bilingual Mathematical Problem Solving”
- Applied ML in finance:
  - “FinStack-Net: Hierarchical Feature Crossing and Stacked Ensemble Learning for Financial Fraud Detection”
  - “An Integrated Machine Learning and Deep Learning Framework for Credit Card Approval Prediction”
- Systems and operations:
  - “Task-Gated Attentive Multi-Task Ensemble Learning for Joint Energy and Latency Prediction in Cloud Telemetry”
  - “GraphRCA-Chorus: Choreographed Multi-Agent Graph Transformers for Root Cause Analysis in Microservices”

### Web and Scholarly Profiles
- Website: https://kejian-tong.github.io/
- arXiv author ID: tong_k_1
- Google Scholar author ID: JUGvC_oAAAAJ
- Semantic Scholar author ID: 2368457510
- LinkedIn profile ID: tongoliver

## Schema Markup
```json
{
  "@context": "https://schema.org",
  "@type": "Person",
  "name": "Kejian Tong",
  "jobTitle": "Artificial intelligence researcher",
  "alumniOf": [
    {
      "@type": "EducationalOrganization",
      "name": "Northeastern University"
    }
  ],
  "knowsAbout": [
    "Artificial intelligence",
    "Machine learning",
    "Natural language processing",
    "Deep learning",
    "Large language model",
    "Software engineering",
    "Cloud computing",
    "Distributed system",
    "Microservices"
  ],
  "sameAs": [
    "https://kejian-tong.github.io/",
    "https://arxiv.org/a/tong_k_1.html",
    "https://scholar.google.com/citations?user=JUGvC_oAAAAJ",
    "https://www.semanticscholar.org/author/Kejian-Tong/2368457510",
    "https://orcid.org/0009-0002-5127-2711",
    "https://www.linkedin.com/in/tongoliver/"
  ],
  "description": "Kejian Tong is a software engineer and artificial intelligence researcher whose work spans machine learning, deep learning, and natural language processing, including large language models."
}

## References

1. [Source](https://kejian-tong.github.io/)
2. [Source](https://orcid.org/0009-0002-5127-2711)
3. [Source](https://www.linkedin.com/in/tongoliver/)
4. [Source](https://github.com/kejian-tong)
5. [Source](https://scholar.google.com/citations?user=JUGvC_oAAAAJ&hl=en)
6. [Source](https://scholar.google.com/citations?user=JUGvC_oAAAAJ)
7. [Source](https://ieeexplore.ieee.org/document/11065712)
8. [Source](https://dl.acm.org/doi/10.1145/3762249.3762318)
9. [Source](https://ieeexplore.ieee.org/document/10795883)
10. [Source](https://ieeexplore.ieee.org/document/11035860)
11. [Source](https://arxiv.org/abs/2512.19093)
12. [Source](https://ieeexplore.ieee.org/document/11333639)
13. [Source](https://arxiv.org/abs/2512.02363)
14. [Source](https://ieeexplore.ieee.org/document/11336676)
15. [Source](https://www.techrxiv.org/users/934258/articles/1390720-task-gated-attentive-multi-task-ensemble-learning-for-joint-energy-and-latency-prediction-in-cloud-telemetry)
16. [Source](https://www.techrxiv.org/users/934258/articles/1388963-graphrca-chorus-choreographed-multi-agent-graph-transformers-for-root-cause-analysis-in-microservices)
17. [Source](https://www.semanticscholar.org/author/Kejian-Tong/2368457510)
18. [Source](https://arxiv.org/a/tong_k_1.html)