Behzad Golshan

Behzad Golshan

Building LLMs & Unleashing Their Potential

A LITTLE ABOUT ME

Hello random person on the internet! My name is Behzad (بهزاد).
I work on Large Language Models (LLMs), Natural Language Processing (NLP) & Deep Learning. Currently, I’m at Google working on bringing LLM capabilities to different products at Google (e.g., SGE).
Before that, I was part of the NLP Signals team at Twitter working on building Tweet representations. But then Musk happened 🙃. And before Twitter, I was doing NLP research at Megagon.ai.
Besides work, I enjoy making & drinking coffee ☕.studying & playing 🎵.supporting AC Milan ⚽.running 🏃 (if my bad knee allows).sleeping 💤.

THE TODO LIST

This is a list of things (first compiled on July 2022) that I wish to achieve. Just like OKRs, I’d say reaching 70% would be a good 😅. Obviously, the list doesn’t include everything cause some goals are really personal. Also, I’ll keep adding to the list (and occasionally remove from it) cause people change.

  • Don’t die
    • I’ve done a good job so far.
  • Run 🏃 a half-marathon
    • My longest run is 8.3 Miles

  • Visit 30 different countries 🗺️ 
    • I’m at 13 countries (excluding Iran & US)

    • 🇦🇪 🇮🇹 🇫🇮 🇯🇵 🇲🇽 🇨🇷 🇧🇷 🇨🇦 🇳🇱 🇧🇪 🇱🇺 🇵🇹 🇪🇸
  • Get 5K followers on Twitter
    • I’m at 769 followers

  • Visit 20 National Parks 🏞️ in US
    • I’m at 6 national parks

    • Yosemite, Joshua Tree, Rocky Mountain, Haleakalā, Pinnacles, Grand Canyon
  • Train to do 20 pull-ups
    • I’m at 0 pull-ups!

  • Watch an AC Milan 🔴⚫ match at San Siro
  • Run 🏃 a marathon
    • My longest run is 8.3 Miles

  • Aprender a hablar español 
    • If we know each other and you speak Spanish, help me! 😅
    • I’m at 104 out of 1193 Duolingo lessons

  • Write a song
  • Win a best-paper award 
  • Learn to sail ⛵
  • Learn to play the piano 🎹 (relatively well)
  • See the northern lights 
  • Write a (non-technical) book 📚
  • Participate in an Olympic-distance triathlon 🏊 🚴 🏃
             – I need to take proper swimming lessons
  • Give a keynote talk
  • Learn to play guitar 🎸 (just strumming chords is fine)
  • Learn to surf 🏄
  • Run 🏃 a mile under 6 minutes
  • Have coffee ☕ with a notable person I admire (author, musician, etc)
             – If you are Paul McCartney, please agree to this!
  • Learn to harmonize 
  • Learn to work a camera 📷 & make good-quality videos
  • Become a good cook 🧑🏻‍🍳
             – I’m an ok cook, but I’d like to learn more.
  • Play music 🪕 on the street
  • Learn to ski ⛷️
  • Get into rock climbing 🧗🏻
  • Finish a half Ironman triathlon 🏊 🚴 🏃
  • Be a guest on a television 📺 show or a podcast 

LET’S SOCIALIZE

PUBLICATIONS

Xianjing Liu, Behzad Golshan, Kenny Leung, Aman Saini, Vivek Kulkarni, Ali Mollahosseini, Jeff Mo
Sainyam Galhotra, Behzad Golshan, Wang-Chiew Tan
Johannes Bjerva, Nikita Bhutani, Behzad Golshan, Wang-Chiew Tan, Isabelle Augenstein
Sampo: Unsupervised Knowledge Base Construction for Opinions and Implications — AKBC 2020
Nikita Bhutani, Aaron Traylor, Chen Chen, Xiaolan Wang, Behzad Golshan, Wang-Chiew Tan
A. Traylor, C. Chen, B. Golshan, X. Wang, Y. Li, Y. Suhara, J. Li, C. Demiralp, W. Tan
Emu: Enhancing Multilingual Sentence Embeddings with Semantic Specialization — AAAI 2020
Wataru Hirota, Yoshihiko Suhara, Behzad Golshan, Wang-Chiew Tan
B. Golshan, G. Mihaila, C. Chen, J. Engel, Al. Halevy, Y. Suhara, W. Tan, M. Matuschek
Danni Ma, Chen Chen, Behzad Golshan, Wang-Chiew Tan
Behzad Golshan, Evimaria Terzi, Panayiotis Tsaparas
A team-formation algorithm for faultline minimization — Expert Systems with Applications 2019
Sanaz Bahargam, Behzad Golshan, Theodoros Lappas, Evimaria Terzi
Xiaolan Wang, Jiyu Komiya, Yoshihiko Suhara, Aaron Feng, Behzad Golshan, Alon Halevy, Wang-Chiew Tan
Xiaolan Wang, Aaron Feng, Behzad Golshan, Alon Halevy, George Mihaila, Hidekazu Oiwa, Wang-Chiew Tan
A. Asai, S. Evensen, B. Golshan, A. Halevy, V. Li, A. Lopatenko, D. Stepanov, Y. Suhara, W. Tan, Y. Xu
Chen Chen, Behzad Golshan, Alon Y Halevy, Wang-Chiew Tan, AnHai Doan
Behzad Golshan, Evimaria Terzi
Homogeneity in web search results: Diagnosis and mitigation — Transactions on Intelligent Systems and Technology 2017
Rakesh Agrawal, Behzad Golshan, Evangelos E Papalexakis
Data integration: After the teenage years — SIGMOD/PODS (Invited Paper)
Behzad Golshan, Alon Halevy, George Mihaila, Wang-Chiew Tan
Esther Galbrun, Behzad Golshan, Aristides Gionis, Evimaria Terzi
Rakesh Agrawal, Behzad Golshan, Evangelos Papalexakis
Rakesh Agrawal, Behzad Golshan, Evangelos Papalexakis
Overlap Between Google and Bing Web Search Results! Twitter to the Rescue? — COSN (Conference on Online Social Networks) 2015
Rakesh Agrawal, Behzad Golshan, Evangelos Papalexakis
Whither social networks for web search? — KDD 2015 (Industry Track)
Rakesh Agrawal, Behzad Golshan, Evangelos Papalexakis
A study of distinctiveness in web results of two search engines — WWW 2015 Companion (Web Science Track)
Rakesh Agrawal, Behzad Golshan, Evangelos Papalexakis
Profit-maximizing cluster hires — KDD 2014
Behzad Golshan, Theodoros Lappas, Evimaria Terzi
Grouping students in educational settings — KDD 2014
Rakesh Agrawal, Behzad Golshan, Evimaria Terzi
Unveiling Variables in Systems of Linear Equations — SDM 2014
Behzad Golshan, Evimaria Terzi
Rakesh Agrawal, Behzad Golshan, Evimaria Terzi
What do row and column marginals reveal about your dataset? — NeurIPS 2013
Behzad Golshan, John Byers, Evimaria Terzi
A framework for evaluating the smoothness of data-mining results — PKDD 2012
Gaurav Misra, Behzad Golshan, Evimaria Terzi
Sofia Search: a tool for automating related-work search — SIGMOD 2012 (Demo Track)
Behzad Golshan, Theodoros Lappas, Evimaria Terzi

WORK EXPERIENCE

Google

Google Software Engineer (2023 – Now)

I’m working on incorporating Google‘s LLM capabilities into different Google products.

Twitter

Twitter ML Engineer (2022-2023)

I have been working on developing and utilizing NLP signals to improve different products at Twitter.

Megagon Labs

Megagon Labs Research Scientist (2016-2021)

I have been working on different NLP & machine-learning problems. Creating knowledge-bases (KBs) automatically from reviews, boosting performance of NLP models through KBs, and building question-answering (QA) and dialogue systems are a few examples.

One of my favorite projects was creating a dataset of self-reported moments that made people happy.

EDUCATION HISTORY

Boston University

Boston University PhD in Computer Science (2010-2016)

I did my PhD at Massive Data, Algorithms, and Systems (MiDAS) Group under supervision of Prof. Evimaria Terzi. Most of my PhD work has been focused on data-mining methods and combinatorial optimization algorithms for real-world graphs.

University of Tehran

University of Tehran BSc in Computer Engineering (2005-2010)

I love this place! This is where I started, made wonderful friends, and learned a lot about computers, algorithms, and myself.

INTERNSHIPS EXPERIENCES

Aalto University

Aalto University Research Intern (Summer 2015)

Microsoft Research

Microsoft Research Research Intern (Summer 2014)

oDesk (now Upwork)

oDesk (now Upwork) Research Intern (Summer 2013)

Sapienza University

Sapienza University Research Intern (Summer 2012)

Adverplex (now Cogo Labs)

Adverplex (now Cogo Labs) Research Intern (Summer 2011)

VIEW MY LINKEDIN PROFILE

MY LINKEDIN PROFILE