Projects | Mohammad Aghajani Asl

A selection of my research and engineering work in trustworthy AI, agentic AI, Persian NLP, intelligent systems, and large-scale data infrastructure.

Research Projects

Academic and research-oriented projects focused on trustworthy AI, multi-hop reasoning, question answering, and evidence-grounded generation.

FAIR-RAG

An evidence-driven agentic RAG framework that identifies missing information and iteratively refines retrieval for complex multi-hop questions.

FARSIQA

An end-to-end, safety-aware Persian question-answering system built on FAIR-RAG for faithful reasoning over authoritative domain knowledge.

IslamicPCQA

A large-scale Persian dataset and benchmark for multi-hop complex question answering over unstructured Islamic text resources.

HERMES

A human-guided, evidence-grounded multi-agent system supporting the complete scholarly workflow for long-form humanities research and writing.

Applied AI Systems

End-to-end AI systems designed for real-world users, operational constraints, decision support, and intelligent product experiences.

HOOSHYAR

An explainable AI advisory system that evaluates fraud-risk indicators in online classified listings and helps users make safer transaction decisions.

TRENDGEO

A geospatial social-media intelligence platform for detecting, mapping, analyzing, and comparing regional and multilingual trends.

TIZBIN

A social-media user-discovery and behavioral-intelligence platform for profile enrichment, activity analysis, relationship discovery, and model-assisted user insights.

Data Engineering & Infrastructure

Large-scale crawling, indexing, dataset construction, model training, and production data infrastructure for Persian-language AI systems.

TELEGRAM_CRAWLER

A scalable and fault-tolerant Telegram data-collection infrastructure built with Telethon for asynchronous crawling, incremental synchronization, normalization, and searchable indexing.

TWITTER_CRAWLER

A scalable and recoverable social-data ingestion infrastructure for collecting, normalizing, enriching, and indexing large continuously updated Twitter/X datasets.

Large-scale Persian NLP & Data Infrastructure

An end-to-end data and model infrastructure for constructing Persian-language datasets, training NLP models, and deploying production services for sentiment, topic, and geolocation analysis.