Sara SalamatAI/ML Engineer
5+ Years in AI / ML
Agentic Systems in Capital Markets
10+ Publications in LLM / NLP
About Me .
I started my journey into AI as an engineering student who couldn't stop volunteering for PhD research projects involving data. That curiosity turned into a path, through online courses, electives from the ECE department, and a BSc thesis on reinforcement learning.
After an AI bootcamp and an internship at ParticleB, where I worked on NLP-driven crypto market analysis and built graph neural network models for trading signals, I knew this was what I wanted to do. I pursued an MSc in Computer Engineering with a focus on Information Retrieval and NLP, published my research at top-tier venues including ECIR, CIKM, and the Machine Learning Journal.
Since then I've built AI systems across recommendation and retrieval in academic domain, and financial reasoning, ranging from neural ranking pipelines to fully orchestrated multi-agent platforms. Today at KX, I work on research and development of AI agents in finance. I spend a lot of time learning, following new technologies, and finding ways to connect what's emerging to what's actionable.
Featured Projects .
Some cool stuff I've built
End-to-end ML pipeline that pulls data for a specific conference, creates enriched index for search and lets user ask question about the conference, trends, popular publications, etc.
Author Identity Disambiguation System
A service to intelligently search and disambiguate academic authors from partial or ambiguous input information.
Research .
Publications in IR, NLP, and LLMs
A contrastive neural disentanglement approach for query performance prediction
S Salamat, N Arabzadeh, S Seyedsalehi, A Bigdeli, M Zihayat, E Bagheri
Gender disentangled representation learning in neural rankers
S Seyedsalehi, S Salamat, N Arabzadeh, S Ebrahimi, M Zihayat, E Bagheri
Benchmarking prompt sensitivity in large language models
A Razavi, M Soltangheis, N Arabzadeh, S Salamat, M Zihayat, E Bagheri
exHarmony: Authorship and Citations for Benchmarking the Reviewer Assignment Problem
S Ebrahimi, S Salamat, N Arabzadeh, M Bashari, E Bagheri
RottenReviews: Benchmarking Review Quality with Human and LLM-Based Judgments
S Ebrahimi, S Sadeghian, A Ghorbanpour, N Arabzadeh, S Salamat, M Li, et al.
Building Trustworthy Peer Review Quality Assessment Systems
N Arabzadeh, S Ebrahimi, A Ghorbanpour, S Sadeghian, S Salamat, M Li, et al.
Reviewerly: Modeling the Reviewer Assignment Task as an Information Retrieval Problem
N Arabzadeh, S Ebrahimi, S Salamat, M Bashari, E Bagheri
Neural Disentanglement of Query Difficulty and Semantics
S Salamat, N Arabzadeh, S Seyedsalehi, A Bigdeli, M Zihayat, E Bagheri
Learning Query-Space Document Representations for High-Recall Retrieval
S Salamat, N Arabzadeh, F Zarrinkalam, M Zihayat, E Bagheri
Neural Ad-Hoc Retrieval Meets Open Information Extraction
DT Vo, F Zarrinkalam, B Pham, N Arabzadeh, S Salamat, E Bagheri
Don't Raise Your Voice, Improve Your Argument: Learning to Retrieve Convincing Arguments
S Salamat, N Arabzadeh, A Bigdeli, S Seyedsalehi, M Zihayat, E Bagheri
Query Performance Prediction Using Neural Query Space Proximity
A Bigdeli, S Ebrahimi, N Arabzadeh, S Salamat, S SeyedSalehi, et al.
Text Representation Enrichment Utilizing Graph based Approaches: Stock Market Technical Analysis Case Study
S Salamat, N Tavassoli, B Sabeti, R Fahmi
Blog .
Occasional thoughts on AI, engineering, and things I'm learning
Let's Connect .
I'd love to hear about new challenges, ideas, and opportunities.
©Sara Salamat · 2026