AI Engineer ยท ML Researcher ยท CS Student

Omid Naeej Nejad

Building systems that see, read, and reason
from pixels and tokens to decisions.

Final-year B.Sc. student at the University of Tehran (GPA 3.65/4.0), specializing in deep learning across computer vision, NLP, and generative AI. 1st Place at Hackathon Pol 2026 ยท TOEFL 89.

Omid Naeej Nejad

Who I Am

I am a final-year Computer Science student at the University of Tehran, where I have spent four years building a foundation in AI engineering that spans computer vision, natural language processing, generative models, and reinforcement learning. My academic path has been shaped by a conviction that the most interesting problems sit at the intersection of perception and reasoning โ€” where systems must not just recognize the world but act intelligently within it.

Beyond coursework, I have contributed to applied research in two university labs โ€” co-authoring a survey on object-centric Vision Transformer architectures for semantic scene understanding, and engineering RAG-based chatbot systems and AI-for-finance pipelines at the Applied AI Lab. In industry, I have shipped BI dashboards and data pipelines at a commercial organization, and in June 2026 I led my team to 1st Place at Hackathon Pol, a national AI competition focused on FMCG demand forecasting.

I care deeply about the craft of building AI systems that are both technically rigorous and genuinely useful โ€” models trained thoughtfully, systems architected for production, and research communicated clearly. Outside the lab, I teach: I have served as Chief Teaching Assistant for three courses and TA for six more, and have co-taught AI extracurricular courses at secondary schools as a volunteer.

3.65 GPA / 4.0
9 Courses TA'd
๐Ÿฅ‡ Hackathon Pol 2026
10+ AI / ML Projects

Research Interests

Explainable & Unified Multimodal AI

Building interpretable AI systems that unify perception across text, image, audio, and video modalities โ€” moving toward models that explain their own reasoning.

XAIMultimodalVLMs

Intelligent Multi-Agent Systems

Designing autonomous agents that coordinate, negotiate, and make decisions in complex environments โ€” including hierarchical reinforcement learning and emergent behavior.

Multi-AgentRLAutonomy

Vision Language Models & Scene Understanding

Object-centric representations, Vision Transformers, scene graphs โ€” developing richer semantic understanding of visual environments beyond bounding-box perception.

ViTObject-CentricScene Graphs

LLMs as High-Level Decision Makers

Using large language models as reasoning engines for autonomous vehicles, robotics, and automated systems โ€” grounding world knowledge in real physical decisions.

LLMsRoboticsPlanning

Publications

In Preparation

Object-Centric Vision Transformer Architectures for Semantic Scene Understanding: A Survey

Omid Naeej Nejad et al.

Computer Vision Lab, University of Tehran ยท 2025โ€“2026

A comprehensive survey synthesizing state-of-the-art methods in object detection, semantic segmentation, scene graphs, and ViT-based object-centric scene representations.

Work & Research

Vision Researcher

Computer Vision Lab ยท University of Tehran

Oct 2025 โ€“ Feb 2026 Research
  • Co-authoring a survey on object-centric ViT architectures for semantic scene understanding, synthesising state-of-the-art methods in detection, segmentation, and scene graphs.
  • Conducted systematic literature review across 20+ papers in the object-centric representation learning space.

BI Engineer Part-Time

Strategy & Development Dept. ยท Pardis Sanat Siyare Sabz (GREEN)

May 2025 โ€“ Nov 2025 Industry
  • Designed and deployed interactive BI dashboards surfacing KPIs for C-level strategic decision-making.
  • Investigated and documented modern data management pipelines, evaluating ETL tooling and data warehousing options for enterprise adoption.
  • Collaborated cross-functionally with domain experts to translate business requirements into actionable data products.

AI Researcher & Developer

Applied AI Lab ยท University of Tehran

Nov 2024 โ€“ Oct 2025 Research
  • Engineered and evaluated RAG-based chatbot systems using LangChain and LLM APIs, improving retrieval precision for domain-specific Q&A.
  • Developed AI-driven analytics pipelines for Finance use-cases, applying NLP and supervised learning to extract insights from structured and unstructured data.

AI Product Development Bootcamp

ZAI Bootcamp ยท Azadi Innovation Factory

Jan 2026 Bootcamp
  • Intensive programme on AI product development, MVP design, and business-AI integration โ€” working alongside domain expert mentors on applied problem-solving.

Academic Background

B.Sc. in Computer Science

University of Tehran

Sep 2022 โ€“ Jul 2026 (Expected) GPA: 17.88 / 20  โ‰ˆ  3.65 / 4.0

Focus: AI ยท Deep Learning ยท Computer Vision ยท IR Systems

Graduate Courses

  • Deep Learning with Applications 16.75/20
  • Machine Vision 14.5/20

Perfect Scores

  • Advanced Information Retrieval 20/20
  • Data Mining 20/20
  • Artificial Intelligence 20/20
  • Design & Analysis of Algorithms 20/20
  • Operating Systems 20/20

Core Courses

  • Non-Linear Programming 17.3/20
  • Bio-Computing 19.05/20
  • Probability 17.6/20
  • Linear Algebra 17/20

Selected Work

2025

Attribute-Conditioned Cartoon Face Generation

Generative pipeline combining conditional GANs with a diversity-aware RL reward signal to produce attribute-controlled cartoon faces with high visual variety, reducing mode collapse.

PyTorchGANReinforcement Learning
2025

Mini GPT โ€” Friends Dialogue

GPT-style decoder-only Transformer implemented from scratch and trained on the Friends TV-show dialogue corpus.

PyTorchTransformersCustom Tokenizer
2025

Anomaly Detection for Medical Images

Heterogeneous autoencoder-based anomaly detection model for medical imaging, developed collaboratively with code review and joint evaluation protocols.

PyTorchAutoencoderMedical AI
2025

Image Captioning โ€” Encoder-Decoder + Attention

Image captioning model using ResNet-101 encoder and LSTM decoder with soft-attention mechanism, achieving coherent natural language descriptions of visual scenes.

PyTorchResNet-101LSTMAttention
2025

Speech Emotion Recognition

Multi-modal speech emotion classifier fusing Mel Spectrogram CNN features with HuBERT self-supervised embeddings, evaluated on standard emotion corpora.

PyTorchHuBERTMel SpectrogramCNN
2025

CNN Architectures & Siamese Networks

Reproduced and benchmarked modern CNN architectures for image classification; built a Siamese network for one-shot face verification using contrastive loss.

PyTorchEfficientNet-B0Siamese
2024

Medical Question Summarization via Round-Trip Translation

Data augmentation pipeline using round-trip neural machine translation to improve summarisation quality on a low-resource medical Q&A dataset, evaluated with ROUGE metrics.

PyTorchTransformersNLTK

Technical Expertise

AI / Machine Learning

Deep Learning PyTorch TensorFlow CNNs GANs Autoencoders Reinforcement Learning Scikit-Learn

NLP / LLMs

Transformers RAG Prompt Engineering LangChain LLMs NLTK Hugging Face HuBERT Text Summarization

Computer Vision

Vision Transformers (ViT) TimeSformer Object Detection Semantic Segmentation Face Recognition Image Captioning Torchvision

Programming

Python C++ C Java Assembly

Data Science & Databases

Pandas NumPy Matplotlib Seaborn Jupyter MySQL NoSQL Data Warehousing

Engineering & Tools

Git / GitHub LaTeX Docker Excel ANTLR4 Logisim

Foundations

Linear Algebra Probability & Statistics Non-Linear Optimization Algorithms & Data Structures

Teaching & Mentorship

Chief TA Feb 2025 โ€“ Feb 2026

Managed TA teams, owned curriculum for assignments, and led project design.

Advanced Information Retrieval

Search algorithms, indexing pipelines, probabilistic retrieval models, and large-scale web data processing.

University of Tehran

Bio-Computing

Metaheuristic algorithms for NP-hard problems: genetic algorithms, simulated annealing, swarm intelligence.

University of Tehran

Data Structures & Algorithms

Foundational CS course covering sorting, trees, graphs, and algorithmic analysis with a focus on implementation.

University of Tehran

Teaching Assistant Sep 2024 โ€“ Feb 2026

Designed and graded assignments, quizzes, and provided academic support.

Data Mining
Basic Programming (twice)
Advanced Programming
Operating Systems
Compilers
Volunteer

AI Instructor

Co-taught extracurricular AI courses at secondary schools โ€” Foundations of AI and Prompt Engineering. Co-designed lesson plans with a co-instructor.

Institute for Research in Fundamental Sciences (IFS) Dec 2025 โ€“ Jan 2026

Awards & Certifications

Honors

1st Place

Hackathon Pol 2026

Iran AI Factory ยท Azadi Innovation Factory ยท June 2026

Led a team to victory in a national AI hackathon by building a data-driven forecasting solution for FMCG sell-in data, outperforming all competing teams across machine learning, product design, and business presentation dimensions.

Presentations

Hierarchical Reinforcement Learning

Motivation, structure, and key algorithms โ€” temporal abstraction, options framework, and benefits over flat RL.

AI Course ยท Dec 2025

Web Crawlers and Indexing Pipelines

Architecture of web crawlers, indexing pipelines, and challenges in large-scale web data processing.

Advanced IR Course ยท Dec 2024

PCA Theory & Application in Face Recognition

Foundations of PCA demonstrated through eigenfaces, dimensionality reduction, and feature extraction.

Linear Algebra Course ยท Dec 2023

Certifications

Retrieval Augmented Generation (RAG) with LangChain

DataCamp ยท 2025

Deep Reinforcement Learning in Python

DataCamp ยท 2025

Reinforcement Learning with Gymnasium

DataCamp ยท 2025

Languages

Persian Native
English Proficient ยท TOEFL iBT 89
Italian Beginner

Get in Touch

I am actively looking for research collaborations, internships, and full-time AI/ML engineering roles. If you are working on an interesting problem, I would love to hear about it.