I am a Generative AI engineer and researcher based in Tehran. My work covers domain-specific LLMs: supervised fine-tuning, LoRA/QLoRA, preference alignment, RAG pipelines, and agent systems, plus generative vision models in production.
Generative AI · LLM Systems · AI Research
Ebrahim Mousavi
Generative AI Engineer & Researcher
I build domain-specific language models, fine-tuning and alignment pipelines, RAG systems, AI agents, and multimodal models that can actually be deployed.
- LLMs
- RAG
- AI Agents
- Fine-tuning
- Multimodal AI
- Inference
- PyTorch
About
Engineer first, with a research record
I lead applied R&D on industry collaborations, teach machine learning, and mentor students in deep learning and LLM research. I am looking for industry roles and Ph.D. opportunities in NLP, multimodal learning, and efficient model training.
Expertise
What I build
Systems work across training, retrieval, agents, and serving. Research, teaching, and earlier vision systems are further down the page.
Domain-specific LLM systems
SFT, LoRA/QLoRA, preference alignment (DPO, GRPO), and domain adaptation for models used in research collaborations.
RAG & knowledge systems
Retrieval-augmented generation pipelines paired with domain-specific models and agent workflows.
AI agents
Agent systems built with LangChain, LangGraph, AutoGen, and CrewAI, including document-processing agents.
Efficient AI deployment
Quantization (AWQ/GGUF), 4-bit deployment, vLLM, TensorRT-LLM, Docker, and multi-GPU training with FSDP and DeepSpeed.
Stack
AI engineering stack
Generative AI
Transformers
LLMs
LoRA / QLoRA
DPO / GRPO
RAG
LangChain
LangGraph
AutoGen
CrewAI
ML / Deep Learning
PyTorch
TensorFlow
Scikit-learn
NumPy
OpenCV
Computer Vision
Diffusion Models
Stable Diffusion
Facial Analysis
Multi-modal Learning
AI Infrastructure
vLLM
TensorRT-LLM
Quantization (AWQ / GGUF)
FSDP
DeepSpeed
Docker
Multi-GPU training
Engineering
Python
Linux
Git
SQL
Projects
Selected systems and curricula
Open work that shows how I teach, implement, and apply models. Applied production systems are listed under experience.
LLM-Lab
A from-scratch language model implementation.
Focused on training efficiency and the fundamentals of text generation, rather than wrapping an existing framework.
LLMs and AI Agents Course
An open curriculum on generative AI, LLMs, and agent deployments.
Tutorial series covering how generative models and agents are put into use, maintained as open-source course material.
Machine Learning Course
Open course taught to master’s and Ph.D. students at Tarbiat Modares University.
Supervised and unsupervised learning, with practical Python applications, over five semesters under Prof. Gholam Ali Montazer.
Facial Age Estimation
A PyTorch pipeline that predicts age from facial images.
Deep learning computer-vision work on facial analysis, separate from the later LLM systems.
Experience
Industry roles
-
AI Team Lead — LLMs & Agents
Makan System
Aug 2024 — Present
- Lead LLM and agent R&D for research collaborations on domain-specific models, RAG, and efficient deployment.
- SFT, LoRA/QLoRA, alignment (DPO, GRPO), and 4-bit deployment (+22% benchmark gain, −35% hallucination, −60% latency).
-
AI Team Lead
Resana — Tehran, Iran
Feb 2024 — Aug 2024
- Led generative image models and intelligent document processing agents, reaching 94% accuracy in automated structured data extraction from complex documents.
-
Computer Vision Developer
AIHomeDesign — Remote
Sept 2023 — Feb 2024
- Designed and deployed Stable Diffusion pipelines for real-estate image enhancement, including virtual staging and scene transformation.
- Optimized high-resolution diffusion inference for 10,000+ daily image transformations.
-
Deep Learning Researcher and Developer
Baharan Company — Tehran, Iran
Jan 2021 — Aug 2023
- Implemented production computer vision systems for facial recognition, identity verification, and intelligent image analysis.
-
Network Engineering
Freelance / Contract Roles — Iran
Feb 2017 — Mar 2020
- Designed and maintained secure network infrastructures using Cisco (CCNA-level) and Microsoft enterprise systems.
Research
Publications, teaching, and study
Peer-reviewed work, preprints, university teaching, and mentoring. This sits behind the engineering sections on purpose.
Peer-reviewed journal articles
Published · Q1 · 2025
Pourmirzaei, M., Montazer, G.A., Mousavi, E. ATTENDEE: an AffecTive Tutoring system based on facial EmotioN recognition and heaD posE Estimation to personalize e-learning environment.
Published · Q1 · August 2023
Beygi, M., Fallahi, M., Vali, R., Mousavi, E., Saberian, M., Li, J., Barari, A. FELA-DNN framework to predict the seismic bearing capacity of skirted strip footing built on a non-cohesive slope.
Preprints
Preprint · 2024
Amiri, M., Montazer, G.A., Mousavi, E. Educational Customization by Homogenous Grouping of e-Learners based on their Learning Styles.
Preprint · 2021
Pourmirzaei, M., Montazer, G.A., Esmaili, F., Mousavi, E., Karamizadeh, S. How Self-Supervised Learning Can be Used for Fine-Grained Head Pose Estimation?
In preparation
In preparation
ProtBERT-NER: Transformer-based sequence labeling for post-translational modification sites in cancer-associated proteins. Manuscript in preparation from Ph.D. thesis supervision of Sima Abolhassani Khajeh.
Full list on Google Scholar.
Education
Sept 2018 — Sept 2021
M.Sc. in Information Technology
Artificial Intelligence focus · Tarbiat Modares University, Tehran
Feb 2009 — Feb 2015
B.Sc. in Software Engineering
Islamic Azad University, Isfahan
Teaching
Machine Learning Course
Tarbiat Modares University — Tehran
2020 — 2023
AI Instructor & Technical Writer
Freelance, remote
2023 — Present
LLMs and AI Agents Course
Open-source curriculum
2023 — Present
Mentoring
Sima Abolhassani Khajeh
2024 — Present
Ph.D. student, Islamic Azad University, Qeshm International Branch. Supervised a thesis on protein language models and PTM site prediction (ProtBERT-NER), and mentored ML, deep learning, and LLM coursework.
Zahra Tavakoli Rad
2024 — 2025
Computer science student, Dalhousie University, Canada. Mentored M.Sc. thesis research on code-understanding language models, including model design, training workflows, and scholarly writing.
Engineers & students (50+)
2023 — Present
Remote, independent ML, deep learning, and GenAI mentorship through structured programs on project design, experimentation, and reproducible model development.