Portrait
Fatema Tuj Johora Faria
AI Engineer II
Astha.IT
About Me

Greetings! I am Fatema Tuj Johora Faria, and I earned a Bachelor of Science in Computer Science and Engineering from Ahsanullah University of Science and Technology, Dhaka, Bangladesh.

I am currently working as an AI Engineer II at Astha.IT, where I bring around 2.5 years of full stack experience in AI applications. My work focuses on architecting, developing, and deploying production-grade agentic solutions. I have hands-on experience with Microsoft Azure, Amazon Web Services (AWS), and Google Cloud Platform (GCP) for building scalable, robust, and high-performance systems, while also focusing on user-friendly interfaces that make AI interactions intuitive.

I am always happy to collaborate, exchange ideas, and learn about new research directions. Please feel free to reach out to me at fatema.faria142@gmail.com (personal) or ftj.faria@asthait.com (official).
Research Interests

AI Reasoning & Autonomy:

  • LLM Reasoning: Exploring how large language models can achieve reliable, structured, and multi-step reasoning for tackling challenging real-world scenarios.
  • AI Agents: Examining how AI agents can plan, use tools, collaborate, and adapt to orchestrate complex tasks in dynamic environments.
  • Multimodal AI: Investigating how multimodal models can integrate and reason across diverse modalities (text + images) to improve contextual understanding and decision-making.

Foundations of Dependable AI:

  • AI Safety: Developing AI systems that remain safe, robust, and controllable under uncertainty, unexpected inputs, and real-world conditions.
  • AI Security: Defending language models and AI agents against adversarial attacks, prompt injection, and malicious inputs.
  • Trustworthy AI: Exploring AI systems that promote fairness, explainability, and reliability across diverse environments.

Domain-Specific AI Applications:

  • AI for Databases: Developing AI systems that translate natural language into SQL, reason over database schemas, and enable more intuitive access to structured data across diverse use cases.
  • AI for Social Good: Developing accessible and scalable AI solutions that address real-world societal challenges and create meaningful impact.
Curriculum Vitae
Education (view details )
  • Ahsanullah University of Science and Technology
    Ahsanullah University of Science and Technology
    2019 - 2023
    Bachelor of Science in Computer Science and Engineering
  • National Ideal College
    National Ideal College
    2015 - 2017
    Higher Secondary Certificate
  • Motijheel Model School and College
    Motijheel Model School and College
    2009 - 2015
    Secondary School Certificate
Selected Publications (view details )
CLAIR-Fin: An Adversarial Multi-Agent Framework for Claim-Level Verification and Adaptive Debate in Cross-Modal Financial QA
Selected Under Review
CLAIR-Fin: An Adversarial Multi-Agent Framework for Claim-Level Verification and Adaptive Debate in Cross-Modal Financial QA

Fatema Tuj Johora Faria, Mukaffi Bin Moin, Jubayer Al Mahmud, M. F. Mridha, Md. Alam Hossain

arXiv preprint

Research Area Multi-Agent SystemsMultimodal AIRetrieval-Augmented GenerationTrustworthy AIFinancial Question Answering
CLAIR-Fin: An Adversarial Multi-Agent Framework for Claim-Level Verification and Adaptive Debate in Cross-Modal Financial QA
Selected Under Review
CLAIR-Fin: An Adversarial Multi-Agent Framework for Claim-Level Verification and Adaptive Debate in Cross-Modal Financial QA

Fatema Tuj Johora Faria, Mukaffi Bin Moin, Jubayer Al Mahmud, M. F. Mridha, Md. Alam Hossain

arXiv preprint

Research Area Multi-Agent SystemsMultimodal AIRetrieval-Augmented GenerationTrustworthy AIFinancial Question Answering
TeachMateGPT: A Multi-Agent Knowledge-Grounded Framework for Pedagogical Assessment Generation from Science Curriculum Materials
Selected Under Review
TeachMateGPT: A Multi-Agent Knowledge-Grounded Framework for Pedagogical Assessment Generation from Science Curriculum Materials

Fatema Tuj Johora Faria, Mukaffi Bin Moin, M. F. Mridha, Jubayer Al Mahmud

arXiv preprint

Research Area Multi-Agent SystemsRetrieval-Augmented GenerationAI for EducationEducational AssessmentTrustworthy AI
TeachMateGPT: A Multi-Agent Knowledge-Grounded Framework for Pedagogical Assessment Generation from Science Curriculum Materials
Selected Under Review
TeachMateGPT: A Multi-Agent Knowledge-Grounded Framework for Pedagogical Assessment Generation from Science Curriculum Materials

Fatema Tuj Johora Faria, Mukaffi Bin Moin, M. F. Mridha, Jubayer Al Mahmud

arXiv preprint

Research Area Multi-Agent SystemsRetrieval-Augmented GenerationAI for EducationEducational AssessmentTrustworthy AI
ChannelGuard: Safe Models Do Not Compose into Safe Multi-Agent Systems
Selected Under Review
ChannelGuard: Safe Models Do Not Compose into Safe Multi-Agent Systems

Elias Hossain, Md Mehedi Hasan Nipu, Fatema Tuj Johora Faria, Tasfia Nuzhat Ornee, Maleeha Sheikh

arXiv preprint

Research Area AI SecurityMulti-Agent SystemsPrompt InjectionAdversarial AttacksTrustworthy AI
ChannelGuard: Safe Models Do Not Compose into Safe Multi-Agent Systems
Selected Under Review
ChannelGuard: Safe Models Do Not Compose into Safe Multi-Agent Systems

Elias Hossain, Md Mehedi Hasan Nipu, Fatema Tuj Johora Faria, Tasfia Nuzhat Ornee, Maleeha Sheikh

arXiv preprint

Research Area AI SecurityMulti-Agent SystemsPrompt InjectionAdversarial AttacksTrustworthy AI
Guiding Language Models to Be More Empathetic: Culturally Sensitive Mental Health Advice Generation Through Human-LLM Collaboration
Selected Under Review
Guiding Language Models to Be More Empathetic: Culturally Sensitive Mental Health Advice Generation Through Human-LLM Collaboration

Fatema Tuj Johora Faria, Mukaffi Bin Moin, Md. Mahfuzur Rahman, Khan Md Hasib, Jubayer Al Mahmud, M. F. Mridha

arXiv preprint

Research Area AI for Mental HealthLarge Language ModelsLow-Resource LanguagesCulturally Aware AITrustworthy AI
Guiding Language Models to Be More Empathetic: Culturally Sensitive Mental Health Advice Generation Through Human-LLM Collaboration
Selected Under Review
Guiding Language Models to Be More Empathetic: Culturally Sensitive Mental Health Advice Generation Through Human-LLM Collaboration

Fatema Tuj Johora Faria, Mukaffi Bin Moin, Md. Mahfuzur Rahman, Khan Md Hasib, Jubayer Al Mahmud, M. F. Mridha

arXiv preprint

Research Area AI for Mental HealthLarge Language ModelsLow-Resource LanguagesCulturally Aware AITrustworthy AI
All publications
Technical Skills
Programming Languages Python (NumPy, SciPy, Matplotlib, Pandas, Seaborn)JavaC++
Web Development JavaScriptTypeScriptTailwind CSSFastAPIReact
Databases MySQLPostgreSQLMongoDB
Deep Learning Frameworks TensorFlowKerasPyTorch
LLM Application Frameworks LangChainLangGraphLlamaIndexLlamaAgents
Vector Databases AlloyDB for PostgreSQL (pgvector)ChromaDBFAISSMilvus
Cloud Services (Azure) Azure OpenAIAzure SQL DatabaseAzure Container RegistryAzure App ServiceAzure Blob StorageAzure BoardsAzure FunctionsAzure Bot Service
Cloud Services (AWS) Amazon ECRAWS App RunnerAmazon EC2Amazon S3Amazon SES
Cloud Services (GCP) Google Cloud StorageGoogle App EngineGoogle Compute Engine
Others DockerGitHubOpenCVMicrosoft Bot FrameworkWebSocketJira
News (view details )
2026
2025