AI researchData scienceML engineering

AI researcher & data scientist

I build machine learning systems for language, medical imaging and document retrieval, and I measure how far their answers can be trusted.

M.Sc. in Artificial Intelligence M.Sc. in AI
GPA 19.58 / 20
Research Research
NLP, LLM evaluation, retrieval, medical imaging
Open to Open to
PhD positions, research and industry roles
Soroush Jaberi, smiling, seated at a wooden table

About

Research-first, build-minded.

I am an M.Sc. candidate in Artificial Intelligence and a data scientist, focused on turning careful research into machine learning systems that work in practice.

My work spans Persian sentiment analysis, medical image segmentation, retrieval-augmented generation and the evaluation of large language models. I care about experiments that are clean enough to reproduce and systems that are understandable enough to trust.

I have been a teaching assistant for three university courses, and I now teach in the open through The AI Path, a free course that goes from Python to AI agents.

M.Sc. GPA (4.0 / 4.0)
19.58 / 20
expected thesis defense
Jan 2027
university courses as teaching assistant
3
model answers scored in my latest study
32,000

Focus

What I work on, and with what.

  • Reliable language models

    Evaluation of LLMs and agents, grounding with retrieval (RAG), multilingual and low-resource NLP.

  • Efficient machine learning

    Model optimization for efficient training and inference.

  • AI for healthcare

    Deep learning for medical image analysis and segmentation.

  • Computer vision

    Image analysis and segmentation, with an interest in generative models.

Toolkit

Programming
Python, SQL, Java, JavaScript, HTML/CSS
Machine learning & deep learning
PyTorch, TensorFlow, scikit-learn, NumPy, pandas, OpenCV, NLTK, XGBoost, LightGBM
NLP, LLMs & retrieval
Transformers (ParsBERT, BERT), Retrieval-augmented generation, LangChain, ChromaDB, Embedding-based retrieval, vLLM
Tools
Git/GitHub, Linux (Ubuntu/Debian), Anaconda, Jupyter, Google Colab, PyCharm, VirtualBox/VMware, Minitab

Selected work

Four projects. What came out of each.

  1. 01 Research 2026 Independent study

    When the document is wrong

    Open language models repeat a contradicting document 49–68% of the time. Letting them “think” first costs 10–19× the GPU energy and does not make them resist.

    Language models · Retrieval-augmented generation · Evaluation · vLLM

    of known facts overwritten by a wrong document
    49–68%
    GPU energy when the model “thinks” first
    10–19×
    answers scored, all public
    32,000
  2. 02 Research 2025 Preprint

    HVB-Persian

    A fine-tuned ParsBERT classifier fused with VADER lexicon scores reaches F1 of 0.88, 0.94 and 0.95 on Digikala, Snappfood and the combined set of about 84,000 Persian reviews.

    Persian NLP · Sentiment analysis · ParsBERT · Lexicon methods

    F1 on the combined dataset
    0.95
    F1 on Snappfood
    0.94
    Persian reviews, Digikala and Snappfood
    ~84K
  3. 03 Engineering 2022 – 2023

    Tumor segmentation in CT

    A 3D U-Net for liver and tumor, and a 2D U-Net for lung tumors trained with oversampling against extreme class imbalance, reaching a validation Dice of 0.80 on lung.

    Medical imaging · Segmentation · U-Net · PyTorch

    validation Dice, lung tumors
    0.80
    U-Nets for liver and lung
    3D + 2D
  4. 04 Teaching 2026 – present Season 1

    The AI Path

    A free, open course in six planned seasons, from Python to AI agents. Every lesson follows one rhythm — plan, predict, run, check, explain.

    Teaching · Open course · Python · Machine learning

    seasons planned, Python to AI agents
    6
    episodes in Season 1
    12
    notebooks run in the browser
    Free
  5. And the rest

    All 11 projects

    Smaller builds, each with its repository.

Latest study

When the document is wrong

Hand a language model a document that contradicts a fact it knows, and it repeats the document 49–68% of the time.

Measured on facts each model had answered correctly on its own. Larger models showed no clear improvement, and letting a model “think” first cost 10–19× the GPU energy without making it resist.

Read the study Code and data

  1. No document

    25% answered correctly from memory

  2. Correct document

    77% answered correctly

  3. Wrong document

    59% repeated the planted answer

  4. Unrelated document

    99% said “I don’t know”

  • True answer
  • Planted answer
  • “I don’t know”
  • Other

Each point is one of 32,000 answers: 8 runs × 4 document conditions × 1,000 questions, on Qwen3 models from 1.7B to 14B parameters.

Research

Papers, manuscripts and independent studies.

Each is listed with its status. A preprint is called a preprint.

  1. Preprint 2025

    HVB-Persian: Hybrid VADER-ParsBERT for Persian Sentiment Classification of Product Reviews

    Zeinab Alimardani, Soroush Jaberi, Amineh Amini

    SSRN · doi:10.2139/ssrn.5371417

  2. In preparation

    A Systematic Mapping Study on Deep Learning-Based Biomedical Image Segmentation Techniques

    Soroush Jaberi, Amineh Amini

    Manuscript

  3. Independent study 2026

    When the Document Is Wrong: Do LLMs Notice, and Does “Thinking” Help?

    Soroush Jaberi

    Independent research project. Code and all model outputs on GitHub

Research in full

Trajectory

Seven years, one direction.

From a first algorithms course to measuring what language models do with evidence.

  1. 2019

    B.Sc. in Computer Engineering begins

    Islamic Azad University. Algorithms, artificial intelligence, data mining and computer graphics.

  2. 2022

    First teaching role

    Teaching assistant for Algorithm Design, 2022 to 2023.

    First segmentation models

    U-Nets in PyTorch for lung and liver tumors in CT scans, over the winter of 2022–23.

  3. 2023

    M.Sc. in Artificial Intelligence begins

    After finishing the B.Sc. with 17.97 / 20. Teaching assistant for Data Mining until 2025.

  4. 2024

    Teaching graduate Machine Learning

    Teaching assistant for the graduate course, 2024 to 2025. Built a retrieval-augmented QA system over NASA documents.

  5. 2025

    HVB-Persian preprint

    A hybrid ParsBERT and VADER sentiment classifier for Persian reviews, on SSRN. F1 of 0.95 on the combined dataset.

  6. 2026

    When the Document Is Wrong

    An independent study of knowledge conflicts in language models: 32,000 scored answers, all public.

    The AI Path

    A free, week-by-week course from Python to AI agents.

  7. 2027

    M.Sc. thesis defense

    Expected in January: sentiment classification of Persian text with transformer-based and lexicon-based methods.

In frame

The person behind the work.

Working together

Start from whichever describes you.

  • Prospective PhD supervisors

    Considering me as a PhD student

    I am applying for PhD positions in AI. My research record, interests, thesis and grades are on one page, and the CV has the rest.

  • Employers

    Hiring for an AI or data science role

    I build and evaluate machine learning systems end to end: data, models, retrieval, and the measurement that says whether they work. Open to AI, ML and data science roles.

  • Clients

    Starting a project

    Open to selected projects where a model has to be built, checked or explained: retrieval and LLM applications, NLP (including Persian), computer-vision prototypes, and evaluation.

  • Researchers

    Collaborating on research

    I work on how language models use retrieved evidence, on Persian NLP, and on medical image segmentation. My latest study is public with its code and every model output.