Alireza Salemi
PhD Candidate · UMass Amherst
🎓 Google PhD Fellow
🏆 SIGIR Best Paper ×2
PhD Candidate · University of Massachusetts Amherst
Center for Intelligent Information Retrieval · Advisor: Hamed Zamani
🎓 Google PhD Fellow · 🏆 SIGIR Best Paper ×2
I build systems at the intersection of information retrieval and large language models, focusing on three areas: making LLMs retrieve and use external knowledge more reliably (RAG), adapting LLMs to individual users' history and preferences (personalization), and composing LLMs into agentic systems for multi-step reasoning and autonomous information seeking. A common thread is grounding model behavior in evidence rather than parametric memory. My work has appeared at ACL ('24–'26), SIGIR ('23–'25), EMNLP ('21, '25), ICTIR ('23, '25), ICLR ('26), and WWW ('26).
Optimizing and evaluating retrieval pipelines that make LLMs smarter.
Tailoring LLMs to individual users through benchmarks, RL, self-training, and test-time scaling.
Multi-agent systems for autonomous information discovery and complex reasoning.
Benchmarking and improving LLM memory over millions of tokens of personal context.
Evaluating Retrieval Quality in Retrieval-Augmented Generation
2024A Symmetric Dual Encoding Dense Retrieval Framework for Knowledge-Intensive Visual Question Answering
2023Competitive fellowship recognizing outstanding PhD students in computer science
2025University of Massachusetts Amherst, Manning CICS
2025Manning College of Information and Computer Sciences, UMass Amherst
2023University of Tehran — BSc graduation
2021