Sahel Sharifymoghaddam

PhD student
Computer Science, @UWaterloo

[Sahel Sharifymoghaddam @Sahel_Sharify]

[3/3] Big thanks to @zijian42chen for presenting, @xueguang_ma and the BrowseComp-Plus team for the dataset, and @lintool for the guidance throughout.
📄 https://t.co/8vmSyvGmV7
💻 https://t.co/rzRlBko7sH
🖼️https://t.co/SGw16TMHay

[Sahel Sharifymoghaddam @Sahel_Sharify]

[1/3] Our agentic search work is at #ACL2026! 🎉 I can't make it to San Diego in person, but luckily @zijian42chen is presenting the poster — go say hi!
The one-line takeaway: 𝑹𝒆𝒓𝒂𝒏𝒌 𝒃𝒆𝒇𝒐𝒓𝒆 𝒚𝒐𝒖 𝒓𝒆𝒂𝒔𝒐𝒏. 🧵

[Sahel Sharifymoghaddam @Sahel_Sharify]

[2/3] Thinking more isn't always how LLM search agents should spend their tokens. A light reranking step beats extra reasoning — better accuracy, faster runs, and lower total token cost (measured with 𝒆𝒇𝒇𝒆𝒄𝒕𝒊𝒗𝒆 𝒕𝒐𝒌𝒆𝒏 𝒄𝒐𝒔𝒕).

[Sahel Sharifymoghaddam @Sahel_Sharify]

🙌 Huge thanks to @zijian42chen, @xueguang_ma and the rest of the BrowseComp-Plus team for their valuable dataset, and to my supervisor @lintool for all the guidance along the way.

Would love to hear your thoughts and feedback on this work 👀

[Sahel Sharifymoghaddam @Sahel_Sharify]

🤔 How should LLM search agents spend their reasoning tokens?
❌ Thinking more isn’t always the answer ...
✅ 𝑹𝒆𝒓𝒂𝒏𝒌𝒊𝒏𝒈 often is.
🧵 New paper: https://t.co/oZDWTQ0Z2N

[Sahel Sharifymoghaddam @Sahel_Sharify]

Reranking before reasoning can significantly reduce token usage in deep search agents. We introduce 𝒆𝒇𝒇𝒆𝒄𝒕𝒊𝒗𝒆 𝒕𝒐𝒌𝒆𝒏 𝒄𝒐𝒔𝒕 to measure the true cost tradeoff across different settings.

[Lily Ge @lilyjge]

1/7 Exciting news! Our new paper is out: Lighting the Way for BRIGHT: Reproducible Baselines with Anserini, Pyserini, and RankLLM
📄Paper: https://t.co/yjYAJPfzlu

It’s about retrieval, RAG, and why a twist on BM25 might matter more than you think. 🧵

[Sahel Sharifymoghaddam @Sahel_Sharify]

Thrilled to be heading to #SIGIR2025 with 𝗥𝗮𝗻𝗸𝗟𝗟𝗠! I’ll also be stepping in for a few other exciting projects from the #UniversityOfWaterloo team.
If you’ll be there, let’s chat! 👋
📌 𝗕𝗼𝗼𝗸𝗺𝗮𝗿𝗸 𝗼𝗳 𝗮𝗹𝗹 𝗨𝗪 𝗽𝗮𝗽𝗲𝗿𝘀 available in the images below:

[Sahel Sharifymoghaddam @Sahel_Sharify]

@_reachsumit Great to see RankLLM getting visibility — thanks @_reachsumit for the shoutout!
Check out more details in here: https://t.co/lSTuFlSYYN

[Sahel Sharifymoghaddam @Sahel_Sharify]

🧵1/5
🚨 Excited to announce our SIGIR 2025 resource paper: "𝙍𝙖𝙣𝙠𝙇𝙇𝙈: 𝘼 𝙋𝙮𝙩𝙝𝙤𝙣 𝙋𝙖𝙘𝙠𝙖𝙜𝙚 𝙛𝙤𝙧 𝙍𝙚𝙧𝙖𝙣𝙠𝙞𝙣𝙜 𝙬𝙞𝙩𝙝 𝙇𝙇𝙈𝙨" is now on arXiv! 🎉
📄 Paper: https://t.co/T7IN1iWPzi
💻 Code: https://t.co/O42F2lRCNN
📦 PyPI: https://t.co/7WooEru6Ex

[Sahel Sharifymoghaddam @Sahel_Sharify]

5/5
We hope RankLLM becomes a go-to community resource for reproducible LLM-based ranking research. Check it out, give it a ⭐, contribute, and let us know what you build with it!
📄 https://t.co/T7IN1iXnoQ
🔗 https://t.co/O42F2lSaDl
#SIGIR2025

[Sahel Sharifymoghaddam @Sahel_Sharify]

🧵1/5
🚨 Excited to announce our SIGIR 2025 resource paper: "𝙍𝙖𝙣𝙠𝙇𝙇𝙈: 𝘼 𝙋𝙮𝙩𝙝𝙤𝙣 𝙋𝙖𝙘𝙠𝙖𝙜𝙚 𝙛𝙤𝙧 𝙍𝙚𝙧𝙖𝙣𝙠𝙞𝙣𝙜 𝙬𝙞𝙩𝙝 𝙇𝙇𝙈𝙨" is now on arXiv! 🎉
📄 Paper: https://t.co/T7IN1iWPzi
💻 Code: https://t.co/O42F2lRCNN
📦 PyPI: https://t.co/7WooEru6Ex

[Sahel Sharifymoghaddam @Sahel_Sharify]

4/5
Huge thanks to my amazing collaborators — @rpradeep42, Andre Slavescu, Ryan Nguyen, Andrew Xu, Zijian Chen, Yilin Zhang, Yidi Chen, Jasper Xian, and @lintool! 🙌

🙏 We are also incredibly grateful to all the generous open-source contributors who made this possible!

[Sahel Sharifymoghaddam @Sahel_Sharify]

💻 Code: https://t.co/4sBoK4VUJe
🧠 Work done in collaboration with @Ushivani3 , @WenhuChen , and @lintool .

[Sahel Sharifymoghaddam @Sahel_Sharify]

Check out our latest paper "𝗨𝗻𝗶𝗥𝗔𝗚: 𝗨𝗻𝗶𝘃𝗲𝗿𝘀𝗮𝗹 𝗥𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 𝗔𝘂𝗴𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗟𝗮𝗿𝗴𝗲 𝗩𝗶𝘀𝗶𝗼𝗻 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹𝘀"! #NAACL2025
📄 https://t.co/wxpwnQb0j7

[Sahel Sharifymoghaddam @Sahel_Sharify]

🎉More explainable diagnostics for LLM battle outcomes — now at scale!

Work done with @UWaterloo folks @Ushivani3, @beirmug, @rpradeep42 and @lintool!

Check out full details in our arXiv preprint:
🔗 https://t.co/WK9cimDcEz

[Sahel Sharifymoghaddam @Sahel_Sharify]

Voters in LLM arenas choose battle winners, but don’t always explain!
📚 Building on a recent work from our group, 𝗧𝗵𝗲 𝗚𝗿𝗲𝗮𝘁 𝗡𝘂𝗴𝗴𝗲𝘁 𝗥𝗲𝗰𝗮𝗹𝗹, we bring nugget evaluation to 𝟳𝗞 battles from the 𝗦𝗲𝗮𝗿𝗰𝗵𝗔𝗿𝗲𝗻𝗮 𝗱𝗮𝘁𝗮𝘀𝗲𝘁!

[Sahel Sharifymoghaddam @Sahel_Sharify]

🔎 We also extend to 𝗾𝘂𝗲𝗿𝘆 𝘁𝘆𝗽𝗲 𝗯𝗿𝗲𝗮𝗸𝗱𝗼𝘄𝗻𝘀, and 𝗺𝘂𝗹𝘁𝗶𝗹𝗶𝗻𝗴𝘂𝗮𝗹 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀 identifying where 𝗽𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝗶𝗻𝘃𝗲𝗿𝘀𝗶𝗼𝗻𝘀 arise (e.g., ambiguous or incomplete queries).