Accelerating R&D by connecting the dots between academia and industry
16 March 2026 | 7 min read
Tatiana Torset via Getty
After recognizing the limitations of both bench chemistry and solving for the Schrödinger Equation, Timur Madzhidov embraced AI as the fastest way to answer key chemical questions.
Driven by curiosity about why things happen the way they do, Timur Madzhidov enjoyed a rich and varied academic career that evolved from experimental work to quantum chemistry, and finally to establishing Russia’s first chemoinformatic lab.
Today, as Senior Product Manager for Reaxys, he’s helping develop products that aim to satisfy the curiosity of others – speedily and at scale. Yet, he remains deeply connected to academia. In fact, while on the job, he was awarded his Doctor of Science from the University of Strasbourg. We spoke with Timur about his winding path and how he is always looking to connect academic discovery with solving real-world problems.
How do you describe your job to someone without a technical background?
I say that I apply artificial intelligence technologies to the chemistry field. If they want more details, I explain that it’s like teaching a computer to understand chemistry so it can answer questions or provide insights. When I was in academia, I often taught computers myself. Now at Elsevier, I lead a team of professionals who do this work.
How did you develop an interest in combining chemistry with data science?
It started early on in a way. When I was seven, I told my mom I wanted to be a scientist because I was always curious about how the world works and why things happen the way they do. I had obsessions with medicine, then physics, and finally chemistry. Since chemistry was my last obsession after finishing school, I decided to focus on it when I entered university.
My first experience synthesizing compounds was eye-opening. People think chemists just randomly mix compounds hoping for something useful, but that’s never the case. Before mixing anything, you need to understand everything about it – the goal, the hypothesis, the synthesis plan, safety details. You use databases to find conditions, solvents, whether it’s combustible or explosive. You gather all this information before you actually mix anything.
But the most tedious part for me was purification – removing impurities to isolate your desired compound. Not only did I find this part boring, but the unpredictability also frustrated me. I’d follow protocols perfectly, yet sometimes compounds that could stay in a solution at high concentrations without precipitating couldn’t be dissolved again, or reactions wouldn’t work as expected. At that time, it felt to me more like an art than a science. When I asked my professor why, he’d say, “That's just how chemistry is. You can spend your whole life figuring it out, or you can just focus on getting what you need.”
That frustration pushed me toward quantum chemistry – using mathematical algorithms to understand chemical reactivity by solving the Schrödinger equation. Although it’s nearly impossible to solve it exactly, with some approximations, you can do so, and these approximations help you understand chemical systems. I enjoyed this much more than experimental work because I could then grasp why things happened.
Yet, you didn’t stick with quantum chemistry after getting your PhD
Quantum chemistry has its own limitations. Even simple answers can take months of calculations, whereas an experiment might achieve the same result in a day. Additionally, quantum chemistry works well in the gas phase, but most reactions occur in liquids, where quantum chemical calculations become overwhelmingly complex.
This led me to chemoinformatics and AI applications in chemistry.
Instead of solving the Schrödinger equation, we can use neural networks or machine learning algorithms that find patterns in data – essentially mimicking how chemists think – to answer chemical questions quickly.
Timur Madzhidov, PhD
Senior Product Manager for Reaxys at Elsevier
Onward and upward! And now, looking back at your academic career in chemoinformatics, what are you most proud of?
With Professor Alexandre Varnek from the University of Strasbourg, I established the first chemoinformatics laboratory in Russia at a relatively rural but historically renowned university known for organic chemistry. Before this, people worked in computational drug design but mainly identified as drug designers and focused solely on that application. We created a laboratory entirely dedicated to chemoinformatics as a scientific discipline.
Starting from zero, the laboratory expanded to three scientific groups with over 15 researchers. We gained visibility, published in reputable journals, and attended conferences. We established the first master’s program in chemoinformatics, and I’m especially pleased that our students now work worldwide. The last time I visited Paris, I met four former students, all working there at different companies – two of whom were even competing against each other!
And it’s also funny; we didn’t have any scientists in our family, but I finally convinced my sister to finish the master’s program I started. Now, there are two of us.
How did you transition from academia to Elsevier?
In 2015 at a conference, I met Juergen Swienty-Busch, who was the Reaxys director at the time. I said, “You have so much data in Reaxys – why not collaborate? You provide data, we’ll develop something innovative.” Surprisingly, they agreed, and we joined the Reaxys R&D collaboration network.
We developed some impressive solutions using Reaxys data, but I realized we couldn’t share them with the broader chemistry community because the data belonged to Reaxys and our models belonged to us. We had cracked interesting problems that could benefit chemists, but we couldn't deliver the solutions to them – and this worm stayed in my brain for some time.
Finally, a couple of years ago, I proposed to the Reaxys team: “We’ve had many discussions about these solutions we’re developing. Do you want me to finally develop this within Reaxys?” They agreed, and now I am here in Oxford.
Again, you turned a perceived limitation into a new opportunity. What's the biggest challenge in transitioning scientific ideas to real-world products?
It took a year, maybe two, to overcome my academic naivety and become product-oriented [laughs]. Converting scientific prototypes into user-facing products requires a completely different approach. In academia, if you deliver a solution that works 70% of the time, everyone’s happy. But Reaxys users are used to having much higher expectations – closer to 95% accuracy.
A solution must also be self-explanatory and easy to use without complicated documentation. This requires designers and researchers to present features intuitively, which takes time. For one project, we spent eight months just understanding the problem and what users want.
We also need to assist users in recognizing when predictions might be wrong. Even if they can accept some errors, we must help them understand where predictions are trustworthy and where they aren’t. Users trust us and count on us to guide them.
How do you connect academic research with industry needs?
This is truly my favorite part of working at Elsevier. I constantly talk to customers – from large pharmaceutical companies to small universities all over – so I have insights into what they need. I also stay connected with researchers like Alexandre Varnek, Alexei Lapkin, Philippe Schwaller, Kevin Jablonka, Marvin Segler, and Jonathan Goodman.
I love connecting the dots. When a user mentions a problem, I might recall a discussion I had with, say, Professor Lapkin a year earlier. Sometimes, I connect them directly. Other times, if the problem is common across many users, we consider commercializing these innovative academic ideas by bringing them into Reaxys to make them accessible to everyone.
I also enjoy challenging scientists. For example, just yesterday, I told Philippe about a specific problem our users encounter and suggested it aligns with his research. If it proves scientifically feasible, we could develop it for our users, and he could publish a paper. Everyone benefits.
What do you wish people better understood about your work?
There’s a lot to cover! But first, people often misunderstand what AI can and cannot do. They think it’s all-powerful, but it’s not. Currently, users often confuse AI with ChatGPT-like large language models. While AI, especially in chemistry, is much broader and more diverse in the way it works. We constantly need to explain the differences between various AI technologies and what’s actually possible versus marketing hype.
Now that you are comfortably established in the product world, what drives you?
My goal is to create something useful that benefits everyone and makes them happy – that, in turn, makes me happy. The true reward is connecting people with the data and tools they need to advance chemistry more quickly and efficiently.
It’s truly amazing to see a product come to life and be used, and to know that, in part, it’s because of you considering and understanding all these different nitty-gritty details. It’s like you built a house and now watch others enjoy their lives there. With this house now part of your soul and heritage, it’s a wonderful feeling to realize they’re happy living in it, partly thanks to you. And it makes it worth all the time and effort that went into it.