Architecting
Intelligence.
I am Hossein Bahak. A Software Engineer and NLP Researcher. Passionate about Large Language Models, Linux philosophy, and building systems that scale beyond the theoretical.
2025
MSC Graduate01. Biography & Philosophy
Born in 1998, I am a Linux and FOSS enthusiast who believes in the power of open engineering. My journey is defined by bridging the chasm between academic theory and industrial scalability.
During my Master's at the University of Isfahan, I focused on making Large Language Models more reliable by detecting reasoning flaws—a mission that has led to several high-impact publications, including ACL 2024.
I specialize in Python and Django, but my true passion lies in the orchestration of AI: from RAG frameworks to agentic workflows using LangGraph.
Full-Stack Dev
Django • DRF • PHP • SQL
AI Research
NLP • LLMS • RAG • PyTorch
Big Data
Spark • Hadoop • Neo4j
Infrastructure
Linux • Docker • N8N
02. Professional Path
Machine Learning Engineer
IRANICARD
- Building production grade ML pipelines and agentic AI frameworks for automated reasoning, task orchestration, and end to end workflow automation.
Machine Learning Specialist
RADSHID COMPANY
- Developing Agentic AI Chatbots using LangGraph for multi-phase visual and textual reasoning tasks.
- Implemented vision noise reduction and object removal pipelines using PyTorch, ResNet, and Diffusion Models.
- Architected N8N automation pipelines for content generation and data synthesis.
NLP Researcher
BIG DATA GROUP, UNIVERSITY OF ISFAHAN
Focusing on RAG Frameworks and LLM hallucination benchmarking. Led student workshops on Prompt Engineering and model fine-tuning methodologies.
ML Engineer & Back-End Developer
CHEHELSOTUN COMPANY
Integrated Two-Tower Neural Networks for FMCG delivery optimization. Engineered scalable ecommerce backend systems with high-availability search.
03. Academic Contributions
Google Scholar ↗Drug Demand Forecasting, Isfahan Artificial Intelligence Event 2023
Developed AI-driven models to forecast drug demand and optimize pharmaceutical production and distribution. Analyzed time series data on drug usage in Hamedan, Iran. Published after the IAI Event Contest, where our team secured 3rd place.
Evaluating ChatGPT as a QA System: A Comprehensive Analysis
Awarded at the 22nd International Conference of Iranian Cryptology Association. Comparative analysis of ChatGPT internal knowledge versus RAG systems.
04. The Technical Lab
LoRA-to-Ollama
Automated framework for fine-tuning LLMs via LoRA and deploying as optimized GGUF files in local Ollama environments.
Graph Extractor
Scalable system using BERTopic and Neo4j to convert massive PDF documentation into navigable semantic knowledge graphs.
Anomaly Framework
Integrated VLM system using WinCLIP for few-shot defect identification and logical reasoning in industrial quality inspection lines.
EDUCATION
Master of Software Engineering
University of IsfahanBSc in Computer Engineering
Semnan University
Let's Solve the
Impossible.
Open for research collaboration, senior ML engineering roles,
and complex architectural consultations.