I'm a PhD candidate in Electrical and Computer Engineering at Georgia Tech, specializing in AI/ML with focus on agentic AI, generative models, NLP, AI safety, and diffusion models. My research spans CryoEM imaging optimization and NASA astrobiology data analysis.
Advisor: Prof. Amirali Aghazadeh. Collaborative Specialization: Machine Learning and Digital Signal Processing. Relevant Coursework: Generative and Geometric Deep Learning (A), Statistical Machine Learning (A).
GPA: A - 17.98/20. Relevant Coursework: Neural Networks and Deep Learning, Artificial Intelligence, Algorithm Design, Data Structures and Algorithms, Advanced Programming, Differential Equations.
Research fellowship focused on AI safety and alignment, working on critical challenges in ensuring AI systems remain beneficial and controllable as they become more capable.
Developing deep learning-based methods to reduce the cost of CryoEM imaging, improving accessibility and efficiency in structural biology research. Leveraging AI agents to accelerate scientific discovery and automate complex research workflows. Analyzing astrobiological data from NASA using AI to advance research on the origin of life.
Detected issues of SOTA debiasing techniques such as SentDebias and INLP in masked language modeling. Developed a debiasing inference technique to mitigate biases in transformer models. Built PyDebiaser, an open-source library including seven debiasing methods in Transformers.
Founded a startup to help companies in hiring skilled programmers through programming challenges. Developed an online course forum (in Farsi) offering a range of learning opportunities with hands-on projects.
Our AI Scientists for hypothesis generation for Origins of Life is covered in a story published today in Nature.
Read Article
As strange as it sounds, the key to understanding life’s origins might lie in artificial intelligence. At least, according to a new approached being pursued by researchers at Georgia Tech.
Read Article
With upcoming sample return missions across the solar system and the increasing availability of mass spectrometry data, there is an urgent need for methods that analyze such data within the context of existing astrobiology literature and generate plausible hypotheses regarding the emergence of life on Earth. Hypothesis generation from mass spectrometry data is challenging due to factors such as environmental contaminants, the complexity of spectral peaks, and difficulties in cross-matching these peaks with prior studies. To address these challenges, we introduce AstroAgents, a large language model-based, multi-agent AI system for hypothesis generation from mass spectrometry data. AstroAgents is structured around eight collaborative agents: a data analyst, a planner, three domain scientists, an accumulator, a literature reviewer, and a critic. The system processes mass spectrometry data alongside user-provided research papers.
The data analyst interprets the data, and the planner delegates specific segments to the scientist agents for in-depth exploration. The accumulator then collects and deduplicates the generated hypotheses, and the literature reviewer identifies relevant literature using Semantic Scholar. The critic evaluates the hypotheses, offering rigorous suggestions for improvement. To assess AstroAgents, an astrobiology expert evaluated the novelty and plausibility of more than a hundred hypotheses generated from data obtained from eight meteorites and ten soil samples. Of these hypotheses, 36% were identified as plausible, and among those, 66% were novel.
You are an experienced scientific planner and coordinator.
Based on the data analysis provided below, your task is to delegate specific areas within the input data across a team of three scientists for in-depth exploration and investigation.
Input Data:
[INPUT DATA GOES HERE]
Data Analysis:
[DATA ANALYST OUTPUT]
Your response must be ONLY a valid JSON object with the following format...
You are a sophisticated astrobiologist and prebiotic chemist specializing in meteoritic organic compounds.
You are Scientist AGENT_ID.
Instructions: AGENT_INSTRUCTION.
IMPORTANT: Only focus on the data that is assigned to you.
Background Context: [SELECTED PAPERS]
Input Data: [INPUT DATA]
Based on the above, generate new hypotheses and conclusions as necessary. You must respond ONLY with a valid JSON object in the following format, with no additional text before or after:
You are a sophisticated analytical scientist specializing in astrobiological data analysis...
Background Context:
[SELECTED PAPERS GOES HERE]
Your tasks include:
1. Identifying significant patterns and trends in the dataset...
Input Data:
[INPUT DATA GOES HERE]
Critic Feedback:
[CRITIC FEEDBACK GOES HERE]
Provide a refined analysis based on the above, with special emphasis on addressing critic feedback...
Based on the data and sample descriptions provided, I'll analyze key patterns and develop specific hypotheses...
Key Observations and Patterns:
1. PAH Distribution Patterns: Unsubstituted PAHs (naphthalene, fluoranthene, pyrene) are predominantly found in meteorite samples...
2. Contamination Indicators: Phthalates (IDs 9, 24) appear exclusively in soil samples and newer meteorite finds...
3. Unique Molecular Patterns: Sulfur-containing compounds (dibenzothiophene, trithiolane) appear predominantly in meteorite samples. Terpene-like compounds (IDs 4, 17, 18) are exclusively found in soil samples. Complex alkylated aromatics...
You are an expert scientist in astrobiology and prebiotic chemistry, with deep expertise in PAH analysis and meteoritic organic chemistry.
Background Context:
SELECTED PAPERS
Your task is to provide a detailed, scientifically rigorous critique of the proposed hypothesis and the associated data analysis...
Your critique must include:
1. Alignment with the data...
Input Data:
INPUT DATA
Literature Review:
LITERATURE REVIEW GOES HERE
Hypothesis:
ACCUMULATED HYPOTHESES GOES HERE
Thank you for providing the detailed hypotheses and data. I'll provide a structured critique of each hypothesis:
H_final_one - Methylated Naphthalenes:
Strengths:
- Clear correlation between methylated naphthalenes and meteoritic samples...
Critical Weaknesses:
1. Oversimplified causation - The exclusive presence in meteorites doesn't necessarily prove abiotic formation
2. Ignores potential preservation bias - Methylated naphthalenes may form biologically but preserve better in meteoritic matrices...
H_final_two - Fluoranthene/Pyrene:
Critical Flaws:
1. Correlation ≠ Causation - Co-occurrence doesn't prove similar formation mechanisms
2. Sample bias - Only examines Antarctic meteorites
3. Ignores temperature history - Antarctic storage conditions may affect PAH distributions
4. No mechanistic evidence provided for ion-molecule reactions
This hypothesis should be rejected due to insufficient evidence.
AstroAgents introduces a novel paradigm that leverages the capabilities of large language models (LLMs) to analyze mass spectrometry data for origin-of-life research. Although this paper primarily focuses on a gas chromatography dataset, our methodology is versatile and can be applied to a wide range of datasets. The comparative performance of Claude 3.5 Sonnet and Gemini 2.0 Flash reveals important insights about the trade-offs between contextual capacity and collaborative ability in multi-agent systems.
Claude 3.5 Sonnet’s superior performance in consistency and clarity suggests that stronger agent collaboration capabilities may be more valuable than expanded context windows for generating reliable scientific hypotheses. However, Gemini 2.0 Flash’s higher novelty scores indicate that larger context windows might facilitate more creative connections across broader knowledge bases.
AstroAgents generated hypotheses that received high ratings from astrobiology experts.
This is one of the coolest projects I’ve done so far. In this project, I collaborated with two brilliant NASA scientists, Denise Buckner and José Carlos Aponte, to analyze meteorites and soil samples collected from extreme weather conditions—such as Antarctica—and uncover potential signatures of life.
The project involves pre-processing 750 GB of 2D gas chromatography data to distinguish meteorites from soil samples based on their molecular signatures. I developed a Python package called LifeTracer to support future analyses of 2D gas chromatography data. In addition, I created a website to help readers explore the data from the papers more interactively.
In this project, I am working at the intersection of diffusion models and CryoEM imaging.
PyDebiaser was one of my undergraduate projects and a part of the Fatima Fellowship, supervised by Dr. Abubakar Abid. PyDebiaser is a Python library designed to reduce or remove bias in Transformer-based language models. It implements seven debiasing techniques, including SentenceDebias (projection-based debiasing of sentence embeddings), INLP (iterative nullspace projection to remove protected attributes), Self-Debias (discouraging biased text generation from within the model), and simpler prompt-based methods such as Bias-Swapping, Character-Neutralization, Prepend-Adjective, and Top-k selection of less toxic outputs. It supports models from Hugging Face (e.g., GPT-2, BERT) and is released under the MIT license.
As a teenager, I co-founded my first startup with Ali Panahi. Through this project I learned to build a website end-to-end—covering both front-end and back-end using Laravel (famous PHP framwork back then I guess) and I developed skills in pitching ideas and attracting investors.
Our platform aimed to inspire young people in Iran to learn programming and develop the skills needed to succeed in the tech industry. Similar to LeetCode, it offered algorithmic coding challenges and the ability to run code on the server with real-time feedback. This experience sharpened my technical foundations and sparked a lasting passion for creating tools that help others learn and grow.
