• Hi!
    I'm Saber

    An enthusiastic Software Engineer.

    Download CV

  • I am interested in

    AI Compilers, FPGA CAD tools, and hardware programming

About Me

Who am I?

Hi I'm Saber Gholami.I was born in Iran, and received my B.Sc. and M.Sc from K.N. Toosi University of Technology and Tehran Polytechnic University, respectively. I completed my Ph.D. at Concordia University under the supervision of Professor Hovhannes Harutyunyan at 2022.

Currently, I'm working at Tenstorrent as a Senior AI Compiler Engineer working on our MLIR based compiler, known as tt-mlir.

Prior to this, I was working at Huawei Technologies Canada Co., Ltd as a software engineer working on complex FPGA CAD problems as part of the Router team.

My interests includes, but is not limited to: AI Compilers, MLIR, FPGA CAD Tools, Routing, EDA, Algorithmic Graph Theory, Deep learning for graphs, and Social networks analysis.

AI Compilers

FPGA CAD Tool

Deep Learning in Graphs

Graph Theory

My

Education

Concordia University, Montreal, Canada

  • Activities and Research directions: Algorithm design, Broadcasting, ML for graphs, Social networks analysis, Networks, and Graph theory.
  • Supervisor: Professor Hovhannes Harutyunyan
  • GPA: 4.2/4.3

Tehran Polytechnic University, Tehran, Iran

  • Activities and Research directions: Machine Learning, Learning automata, Social networks, Graph coloring, Influence maximization, and Natural language processing.
  • Supervisor: Professor M. R. Meybodi
  • GPA: 17.38/20

K.N. Toosi University of Technology, Tehran, Iran

  • Activities and Research directions: Geographical calculation, Advanced programming, 3D curve fitting.
  • Supervisor: Dr. Amin Nikanjam
  • GPA: 16.13/20
My

Work Experience

Sr. Engineer, AI Compilers

Tenstorrent AI ULC (2025-present)

Contributed to the design and implementation of MLIR-based AI compilers for Tenstorrent hardware, such as blackhole and wormhole using C++20 and Python:

  • Pushed the generality of the compiler and added support in the optimizer for more operators and graphs. This automation resulted in significant reduction in developers' time.
  • Worked on ttnn-jit frontend and made it pip install-able for easier usage.
  • Developed graph-based approaches for generating ttnn IR from a decorated python function to be used in ttnn-jit. This added support for TMs and reduction ops, as well as dynamically decomposing composite ops.

Software Engineer

Huawei Technologies Canada Co., Ltd (2023-2025)

Contributed to the design and implementation of AI compilers tailored for Huawei's hardware platforms.

  • Developed an end-to-end solution integrating MLIR with TVM, leveraging MLIR's optimization passes specifically optimized for Huawei hardware, alongside TVM's auto-scheduling.
  • Designed new MLIR dialects compatible with MLIR's Tosa and TVM's Relax dialects, and implemented lowering passes for these dialects.
  • Led a team of engineers to design and implement an end-to-end testing infrastructure, enabling validation of multiple flows with automated C code inference capabilities using C++, Python, Shell scr.

Engaged in the development of FPGA CAD tools using C++20 for Place and Route team:

  • Conducted thorough research to identify cutting-edge EDA algorithms tailored to support custom architectures, and developed multiple algorithms focusing on core routing, parallel scheduling and filter generation, leading to a notable 12x acceleration in router flow through the implementation of parallel techniques.
  • Designed the complete clock solution for the tool, achieving a significant 16x speed enhancement compared to the previous solution for placing and routing on the clock network.
  • Collaborated closely with the Timing team to pioneer advancements in clock skew optimization and the resolution of hold violations, resulting in a noteworthy 5\% increase in fmax.

Junior Software Engineer

Virtual Reality Lab, K.N.Toosi University (2015-2016)

  • Developed a Java framework for optimizing the movement of virtual cars in curved highways and bridges using Bezier curve fitting methods.

My recent

Publications

Journals

  • Saber Gholami, Hovhannes A. Harutyunyan, "A Note to Non-adaptive Broadcasting", Parallel Processing Letters, 2024. link
  • Saber Gholami, Hovhannes A. Harutyunyan, "HUB-GA: A Heuristic for Universal lists Broadcasting using Genteic Algorithm", Journal of Communications and Networks, 2023. link
  • Saber Gholami, Hovhannes A. Harutyunyan, "Optimal Broadcasting in Fully Connected Trees", Journal of Interconnection Networks, 2023. link
  • Saber Gholami, A.M.Saghiri, S.M.Vahidipour, M.R.Meybodi, "HLA: A Novel Hybrid Model Based on FixedStructure and Variable Structure Learning Automata". Journal of Experimental and Theoretical Artifcial Intelligence, 2023 link

Conferences

  • Saber Gholami, Hovhannes A. Harutyunyan, "Fully-adaptive Model for Broadcasting with Universal Lists", In the 24th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC) 2022. link
  • Saber Gholami, Hovhannes A. Harutyunyan, "Broadcast Graphs with Nodes of Limited Memory", In the 13th International Conference on ​Complex Networks (ComPleNet), May 30 - June 1, 2022. link
  • Saber Gholami, Hovhannes A. Harutyunyan, "A Broadcasting Heuristic for Hypercube of Trees", In 11th Annual Computing and Communication Workshopand Confernce, CCWC 2021, Las Vegas, NV, USA, January 27-30,2021. IEEE, 2021. link
  • Sahar Bakhtar, Saber Gholami, Hovhannes A. Harutyunyan, "A New Metric to Evaluate Local Communities in Social Networks Using Geodesic Distance". The 9th International Conference on Computational Data and Social Networks, Dallas, TX, 2020. link
  • A.Mohammad Ebrahimi, Saber Gholami, Saieede Momtazi, M.R.Meybodi, A.Abdollahzadeh, "Correlation Analysis of Applications' Features: A Case Study on Google Play". The International Conference on Contemporary Issues in Data Science, Iran, 2019. link
Some recent

Talks

Conference talks

  • Fully-adaptive Model for Broadcasting with Universal Lists, in SYNASC 2022. pdf
  • Broadcast Graphs with Nodes of Limited Memory, in CompleNet 2022. pdf
  • A Broadcasting Heuristic for Hypercube of Trees, in CCWC 2021. pdf

University talks

  • Problems Related to Classical and Universal List Broadcasting , Concordia University 2022. pdf
  • Broadcasting with Nodes of Limited Memory, Concordia University 2022. pdf
  • Broadcasting Problem in a Specific Class of Graphs, Concordia University 2021. pdf
  • The Importance of Individuals and Groups in Social Networks, Concordia University 2020. pdf
What interests me?

Research Experience

AI Compilers

MLIR based AI Compilers for hardware accelarators

FPGA CAD

Routing, Placement, Clock network, Timing Analysis

Machine Learning

Supervised and Reinforcement learning.

Graph Theory

Broadcasting, Centrality indices, Graph coloring.

Social Networks

Influence maximization, Community detection.

DL for Graphs

Using Deep Learning techniques in huge graphs.

What have I taught?

Teaching Experience

John Abbott College

  • Foundations of Web Development
    • Lecturer
    • Apr. - May. 2022

Concordia University

K.N. Toosi University of Technology

Cups of coffee
Courses attended
Courses taught
Publications
The list of

My Skills

As a software engineer, you have to have a great set of skills in programming. Here is a list of mine:

Programming Languages

C++, C, Python, Java, Prolog, TCL, VHDL
90%

Production tools

CMake, Git, Jira
80%

Machine Learning

Scikit-learn, NumPy, SciPy, Pandas
60%

Social Networks and Graph Technologies

Networkx, Gephi
95%

Web Technologies

HTML, CSS, Javascript, Bootstrap, Django
75%

Compiler Technologies

MLIR, TVM
70%
What are

My Hobbies

I love traveling, cooking, video games, working out and photography! Oh, and by the way, who doesn't love friends?

Me and Mark

My supervisor at Huawei at the christmas party, ON, CA

Me and Professor Harutyunyan

My last day as a PhD student, QC, CA

Photography

Old port, QC, CA

Photography

Mountroyal sunset, QC, CA

Cooking

Yes, I cook Neapolitan Pizza from scratch

Photography

Sunset in my homecountry, Tehran, Iran.

Get in Touch

Contact

Feel free to contact me 24/7. You can either send me a message here or get in touch with below options.

Follow me on Facebook.

Or on Instagram.

See my profile on LinkedIn and Research Gate.

+1 438 722 9130

Last Update: Jan. 2026