Large Language Models
LLM-based intelligent systems, language understanding and PEFT fine-tuning.
↗RESEARCH • ENGINEERING • SYSTEMS
AI Researcher & Engineer
Building intelligent systems at the intersection of Large Language Models, RAG, Graph Learning, IoT and Robotics.
01 / PROFILE
An AI engineer and machine-learning researcher working across language models, intelligent devices and physical systems.
I design and develop intelligent systems with a focus on Large Language Models, Retrieval-Augmented Generation, Agentic AI, natural language processing, deep learning and intelligent robotics — taking research ideas from R&D to usable products.
My work also includes technical team leadership, interdisciplinary project management, PEFT-based fine-tuning of generative models, and the design of multi-agent AI architectures.
02 / DIRECTIONS
Questions and systems I am currently studying and building.
LLM-based intelligent systems, language understanding and PEFT fine-tuning.
↗Reliable, knowledge-grounded AI systems with semantic retrieval and evaluation.
↗Graph-augmented retrieval architectures for stronger reasoning in language models.
↗Learning representations from graph-structured data.
↗Combining language models with GNNs to improve graph learning.
↗IoT, industrial intelligence and autonomous robotic systems.
↗03 / SELECTED WORK
Selected AI, robotics and embedded systems from research, product development and academic work.
AI / RAG
Architected and built a RAG pipeline for intelligent banking responses, including semantic retrieval, a vector database, intent detection, slot filling and LLM-as-a-Judge evaluation to reduce hallucination.
View project ↗
ROBOTICS / AI
Led software and electronics teams to turn a laboratory prototype into a product: a dental training robot with a RAG/LLM educational assistant and an application–robot–intelligence architecture.
AI / LLM
Designed PEFT fine-tuning for Gemma-family models using LoRA and adapters, with evaluation of catastrophic forgetting and optimization for Persian and domain-specific use.
AI / DEEP LEARNING
Implemented a full Transformer in PyTorch, including self-attention, multi-head attention, positional encoding and feed-forward layers, applied to sentiment analysis.
ROBOTICS / PERCEPTION
Developed LiDAR processing, ICP scan matching and an Extended Kalman Filter for localization of autonomous robots.
View project ↗
IOT / EMBEDDED
A connected filament drying system with temperature, humidity and wireless control considerations.
View project ↗
IOT / SECURITY
An RFID access device with a web application for user and permission management.
View project ↗
AI / ALGORITHMS
A course project using heuristic algorithms to reduce the execution time of a logic game solver.
04 / TOOLKIT
Grouped by the kind of system being built.
05 / PATH
Education, research leadership and product development drawn from the current CV.
Amirkabir University of Technology · Tehran Polytechnic
Master’s student in Computer Engineering — Artificial Intelligence and Robotics, admitted as a distinguished talent student.
Shahid Nazm Bojnourdi Research Center
Directed research and technology activities for more than three years, including research teams, resource planning and scientific collaboration.
OCS intelligent dental-training robot
Led interdisciplinary computer and electrical teams from R&D to an operational prototype, including technical architecture and task allocation.
First specialized AI bootcamp in North Khorasan
Designed the curriculum, taught AI and machine-learning concepts, and managed the training program.
University of Bojnord
Graduated with distinction in six semesters with a GPA of 18.47 / 20, ranking first in the Computer Engineering cohort.
06 / RECOGNITION
Selected academic, research and competition records from the current CV.
Amirkabir University of Technology
GPA 18.47 / 20 · University of Bojnord
Provincial awardee
North Khorasan selection
Specialized artificial intelligence program
Sharif University of Technology
Research document, 2024, available in the project archive.
Paper and patent for the OCS educational robot are currently under review.
Ongoing research on combining language models with GNNs and Graph RAG architectures for improved reasoning.
08 / CV
View the current CV for education, research, projects and professional experience.
09 / CONNECT
Interested in AI, intelligent systems, research collaboration or engineering projects? Feel free to get in touch.
saeedsafaeeshars@gmail.com ↗