Mojtaba Valipour
PHD STUDENT, COMPUTER SCIENCE AND PART-TIME MASTER OF BUSINESS, ENTREPRENEURHIP AND TECHNOLOGY (MBET) STUDENT
Tapping into technology to revolutionize academic research
MATH
Mojtaba Valipour
PHD STUDENT, COMPUTER SCIENCE AND PART-TIME MASTER OF BUSINESS, ENTREPRENEURHIP AND TECHNOLOGY (MBET) STUDENT
Tapping into technology to revolutionize academic research
Over the course of 12 years spent working in academic research, Mojtaba Valipour has learned many things. But the one observation that spurred his award-winning startup venture was this: collaborating on research publications is time-consuming — and a lot harder than it needs to be.
The computer science PhD student, who holds a Master of Science in Artificial Intelligence Robotics and a Bachelor of Science in Computer Hardware Engineering, was recently granted the Entrepreneurial PhD Fellowship for his startup venture, DaneshBaz.
The research writing platform makes collaboration between academic researchers accessible, efficient, easier and scalable.
“To really advance my research,” Mojtaba says, “I need to be able to work and collaborate with a stranger in another part of the world without it having a negative impact on my career. That’s what this platform is designed to solve.”
Disrupting the status quo of scholarly publishing
Traditionally, issues of trust and geography have limited the size of scholars’ collaborator networks. Wary about the risk of partnering with strangers, researchers may tend to be more conservative in their collaborations, potentially hindering scientific progress — and their careers.
The situation is particularly challenging for graduate students and early-career scholars who don’t always have the connections or experience needed to build collaborative relationships.
Through a system of contribution tracking, DaneshBaz enhances ownership and accountability by tagging individual researchers’ contributions to a publication. If a co-author’s work turns out to be plagiarized or not credible, it reduces the risk and consequences for their collaborators.
The platform also disrupts the traditional approach to academic research, which freezes papers after publication, so they are no longer subject to revision. With the introduction of “living papers,” DaneshBaz gives researchers a chance to continue improving upon and collaborating on their publications.
Balancing business and impact in the MBET
As one of seven Waterloo PhD students admitted to the Master of Business, Entrepreneurship and Technology (MBET) through the inaugural Entrepreneurial PhD Fellowship, Mojtaba is completing the three-year part-time program alongside his regular doctoral studies, tuition-free.
Having specialized in deep learning and machine learning and dabbled in the Velocity startup ecosystem, his journey through the MBET program has been an ideal opportunity to step outside his comfort zone and expand his business knowledge.
“My motivation in pursuing entrepreneurship is always about having an impact. It’s a big deal for me as an engineer to have an impact. Money was never something big for me,” Mojtaba says. “But the problem is, you cannot scale things and it’s not a sustainable solution if you are limited by money. So what I’m learning in the MBET is to find the balance between these two.”
The experience of building connections and learning hands-on with like-minded peers has been an eye-opening one, especially seeing how they navigate the business side of their own ventures.
His supervisor, Dr. Ali Ghodsi, who is jointly appointed to the Department of Statistics and Actuarial Science and the Cheriton School of Computer Science, has also played a key role in his support network, providing guidance, advising on research and championing his decision to pursue the MBET.
“I have loved my experience at Waterloo because it didn’t limit me,” says Mojtaba. “I have never felt that anything was holding me back — there is always support. If you want to do more, you can always do more.”
Research that improves research
Just like his business venture, Mojtaba’s doctoral research is aimed at driving improvements in building artificial general intelligence (AGI) — essentially, human-level intelligence — that makes knowledge discovery less expensive and more efficient.
Working in the area of machine learning, his thesis explores two key questions focused on how to build better large language models (LLMs) — the type that run generative AI programs like ChatGPT.
The first question is around expanding the capabilities of LLMs, so they can solve problems outside the scope of what they were designed for. Specifically, he’s investigating how to leverage the power of LLMs to automate the time-consuming process of producing equations from large data points. With AI that quickly finds formulas representing patterns in data, scientists and engineers could dramatically speed up the process of knowledge discovery.
His second question looks at how to adapt LLMs to work as different functions more efficiently and train many different functions simultaneously. As neural networks continue to expand, it gets increasingly difficult and more expensive to retrain them from scratch. Mojtaba’s goal is to make it faster and more cost-effective to improve and expand AI capabilities by training subsets of an LLM, rather than the entire model.
While staying on top of such a fast-moving research area is a challenge, studying at Waterloo has given him the freedom and support to follow where his questions lead — and to explore them outside the classroom. Full-time PhD internships at both Oracle and Noah’s Ark Lab, the AI research center for Huawei Technologies, have been opportunities to build industry connections, enrich his research and apply his learning in the workplace.
As Mojtaba looks ahead to the future and a PhD defense within his reach, he’s still focused on enjoying every minute of his studies, work experience and startup venture.
“I don’t want to rush things. But I am thinking about pursuing my startup full time at some point. It’s important to me to have the freedom to follow my curiosity and explore without limits.”
Disrupting the status quo of scholarly publishing
Traditionally, issues of trust and geography have limited the size of scholars’ collaborator networks. Wary about the risk of partnering with strangers, researchers may tend to be more conservative in their collaborations, potentially hindering scientific progress — and their careers.
The situation is particularly challenging for graduate students and early-career scholars who don’t always have the connections or experience needed to build collaborative relationships.
Through a system of contribution tracking, DaneshBaz enhances ownership and accountability by tagging individual researchers’ contributions to a publication. If a co-author’s work turns out to be plagiarized or not credible, it reduces the risk and consequences for their collaborators.
The platform also disrupts the traditional approach to academic research, which freezes papers after publication, so they are no longer subject to revision. With the introduction of “living papers,” DaneshBaz gives researchers a chance to continue improving upon and collaborating on their publications.
Balancing business and impact in the MBET
As one of seven Waterloo PhD students admitted to the Master of Business, Entrepreneurship and Technology (MBET) through the inaugural Entrepreneurial PhD Fellowship, Mojtaba is completing the three-year part-time program alongside his regular doctoral studies, tuition-free.
Having specialized in deep learning and machine learning and dabbled in the Velocity startup ecosystem, his journey through the MBET program has been an ideal opportunity to step outside his comfort zone and expand his business knowledge.
“My motivation in pursuing entrepreneurship is always about having an impact. It’s a big deal for me as an engineer to have an impact. Money was never something big for me,” Mojtaba says. “But the problem is, you cannot scale things and it’s not a sustainable solution if you are limited by money. So what I’m learning in the MBET is to find the balance between these two.”
The experience of building connections and learning hands-on with like-minded peers has been an eye-opening one, especially seeing how they navigate the business side of their own ventures.
His supervisor, Dr. Ali Ghodsi, who is jointly appointed to the Department of Statistics and Actuarial Science and the Cheriton School of Computer Science, has also played a key role in his support network, providing guidance, advising on research and championing his decision to pursue the MBET.
“I have loved my experience at Waterloo because it didn’t limit me,” says Mojtaba. “I have never felt that anything was holding me back — there is always support. If you want to do more, you can always do more.”
Research that improves research
Just like his business venture, Mojtaba’s doctoral research is aimed at driving improvements in building artificial general intelligence (AGI) — essentially, human-level intelligence — that makes knowledge discovery less expensive and more efficient.
Working in the area of machine learning, his thesis explores two key questions focused on how to build better large language models (LLMs) — the type that run generative AI programs like ChatGPT.
The first question is around expanding the capabilities of LLMs, so they can solve problems outside the scope of what they were designed for. Specifically, he’s investigating how to leverage the power of LLMs to automate the time-consuming process of producing equations from large data points. With AI that quickly finds formulas representing patterns in data, scientists and engineers could dramatically speed up the process of knowledge discovery.
His second question looks at how to adapt LLMs to work as different functions more efficiently and train many different functions simultaneously. As neural networks continue to expand, it gets increasingly difficult and more expensive to retrain them from scratch. Mojtaba’s goal is to make it faster and more cost-effective to improve and expand AI capabilities by training subsets of an LLM, rather than the entire model.
While staying on top of such a fast-moving research area is a challenge, studying at Waterloo has given him the freedom and support to follow where his questions lead — and to explore them outside the classroom. Full-time PhD internships at both Oracle and Noah’s Ark Lab, the AI research center for Huawei Technologies, have been opportunities to build industry connections, enrich his research and apply his learning in the workplace.
As Mojtaba looks ahead to the future and a PhD defense within his reach, he’s still focused on enjoying every minute of his studies, work experience and startup venture.
“I don’t want to rush things. But I am thinking about pursuing my startup full time at some point. It’s important to me to have the freedom to follow my curiosity and explore without limits.”