Manoj Gopalkrishnan and the computer that escaped its box

Leonard Adleman was on a sofa in tennis clothes when Manoj Gopalkrishnan arrived with a definite idea of his future. He would be a theoretical computer scientist. There would be mathematics, physics, chalk, code. His questions were large: how life emerges, how intelligence develops, how complicated behaviour grows out of simpler parts. The setting was relaxed. The consequences would take years to unfold.

Gopalkrishnan recalls being 23 when he walked into his new doctoral advisor’s office. Adleman welcomed him warmly. For the younger researcher, laboratory work with DNA had hardly been the obvious destination of an undergraduate degree in computer science. Yet that destination was waiting. A computer, it turned out, could lead him well beyond the desk on which people usually kept one.

Today, Gopalkrishnan is the founder and CEO of Algorithmic Biologics in Bengaluru. The company builds technology around molecular information processing. To understand the person behind it, however, a company description is a rather narrow doorway. His life also contains theatre groups, comic poetry and an enthusiasm for making things that extends, quite literally, to what he has worn on his feet.

A sofa, tennis shoes, and an unexpected laboratory

He completed his computer science and engineering degree at IIT Kharagpur in May 2003, then pursued a doctorate at the University of Southern California. Adleman advised the dissertation, Theoretical and Experimental Self-Assembly, completed in December 2008. The title contains the crossing: an abstract account of how parts arrange themselves, coupled to the business of getting physical parts to do it.

That crossing gives his career its shape. Self-organization asks how components, following their own interactions, produce a larger pattern. Gopalkrishnan approaches the question with the tools of information science. He wants to understand the algorithms implicit in natural processes and, in the other direction, find principles in those processes that can help design algorithms. The traffic runs both ways.

His institutional route brought him back to India. He joined TIFR’s School of Technology and Computer Science in Mumbai in February 2009, held a visiting mathematics position at Duke in 2010, and received a Ramanujan fellowship that year. Joint appointments connected him to ICTS and TIFR Hyderabad. In June 2016, after an adjunct appointment, he joined IIT Bombay’s electrical engineering department as an associate professor.

The sequence crosses computer science, mathematics and electrical engineering. His research interests cross further still: reaction networks, information geometry, learning and the thermodynamics of information. The recurring problem gives these subjects a common address. How does a physical system handle information, and what would it take to design one that handles it in a useful way?

The molecules get a syllabus

One answer appeared in his work on statistical inference. A 2015 preprint proposed a molecular scheme for computing maximum likelihood estimators for certain mathematical models. That sounds remote until the underlying ambition becomes clear: take a task associated with statistical software and work out a chemical system whose behaviour can perform the calculation.

In 2017, Gopalkrishnan and collaborators including William Poole, Nick Jones, Thomas Ouldridge and Erik Winfree explored Chemical Boltzmann Machines. Their paper described four ways a stochastic chemical reaction network could implement a Boltzmann machine, a neural network model. Randomness in reactions became a computational resource. The work included a computer simulation that classified and generated handwritten digits. Its results belonged to models, mathematics and simulation; the ambition was to give chemical computation a workable design language.

The following year, with Muppirala Viswa Virinchi and Abhishek Behera, he described a reaction network implementation of a generalized expectation-maximization algorithm. In ordinary statistical work, this method helps estimate a model when some information is hidden. Their scheme connected that task to reactions and to the question of how a cell might infer its environment from partial observations.

A 2019 paper with Abhinav Singh, Carsten Wiuf and Behera pursued learning for Hidden Markov Models. Its proposed Chemical Baum-Welch Algorithm had an especially concrete constraint: each reaction changed one molecule of one species into one molecule of another. The authors proved relationships between the fixed points of the chemical scheme and the conventional algorithm, then simulated the network on an example sequence.

The progression matters because it replaces a broad analogy with particular constructions. Saying that chemistry computes is an invitation. Specifying a network, showing what it does and testing the mathematics supplies something another researcher can inspect. Gopalkrishnan’s work lives in that demanding interval between an attractive idea and a description detailed enough to argue with.

The society of late arrivals

While these questions were taking shape, Gopalkrishnan found another use for language. On March 25, 2004, he organized the first meeting of a USC Wodehouse society outside Leavey Library. The comic minutes record an organizer waiting while the people he expected failed to appear on time. Eventually he left for a drink. The delayed messages and arrivals came afterwards.

His account turns the inconvenience into a verse performance, complete with an extravagantly grand narrator, mounting anxiety and a final piece of practical advice about arranging meetings. It is an excellent fate for a disappointing evening. The gathering may have struggled to assemble its members, but the minutes managed to assemble a joke.

Performance has a longer history in his life. He belonged to Banjara at IIT Kharagpur, working with Rashi Bunny, helped found the USC improv troupe Vidushak, and started The Institute of Fun at TIFR. He also spent time in the ensemble of Improv Comedy Mumbai. The institutional names are familiar from his scientific career; the activities give those places a second life.

His other interests include poetry, Ultimate Frisbee and Lindy Hop. He has also made shoes from discarded bicycle tyres, following instructions he found online, and reported completing an 11-kilometre race in them. The detail is wonderfully material. After so much discussion of what matter might be persuaded to do, here is rubber getting a second assignment.

A company built around a search problem

Algorithmic Biologics was founded in 2021. Its premise draws from the same effort to treat molecular systems as information systems. On the company’s account, its engineers work toward chemical logic gates and networks of molecular reactions that can carry out algorithms. Programmable chemistry is an aspiration with a research history behind it, rather than a convenient metaphor invented for a pitch.

The company’s Tapestry technology addresses molecular search. The basic problem is to find samples containing molecules of interest among many samples. Its approach combines samples in planned ways, then uses computation to recover individual information from the measurements. There is a design problem before the measurement and an inference problem after it. Both give algorithms a practical job.

Gopalkrishnan has described the commercial approach as turning models of molecular processes into software, with inference engines delivered through a cloud application. Existing laboratory workflows matter to that approach. A laboratory has equipment, methods and people already doing a job; a useful new system has to make sense within their working day.

It is a considerable change of audience from a mathematical paper. Researchers need a construction they can inspect. Customers need a tool they can use. Training, support and the mechanics of delivery belong to the second task. Gopalkrishnan’s founder role connects the two, carrying ideas developed through fundamental science into an organization responsible for their practical use.

That financing supported plans to develop the product pipeline, build the sales team and expand geographically. The sum is a company milestone, with a date attached. Behind it sits a much longer chronology of doctoral research, faculty work and collaborations. A funding announcement compresses that history into a number; the history explains why the number has something to fund.

“As a boy, I would dream of featuring in the cricket pages.”

Manoj Gopalkrishnan · April 2024

When Algorithmic Biologics appeared in the newspaper in 2024, Gopalkrishnan remembered a boyhood wish to feature in the cricket pages. The sporting route had not worked out, but he welcomed the alternative. His message celebrated his colleagues and thanked the investors who had supported them. The joke made the announcement personal; the attention to the team made it shared.

The teacher returns to the microphone

Teaching provides another view of his appetite for connections. At TIFR in 2015, he taught Abstract Nonsense and its Applications, a category theory course examining applications across physics, computer science and systems theory. The name has comic possibilities; the syllabus had mathematical obligations. Students worked through a textbook and presented assigned papers.

At IIT Bombay, he designed a graduate course on information, statistics and learning and taught undergraduate network theory. Earlier, a public talk called Doctor Logic and Mr. Fantasy used the Ramayana to explain Cantor’s diagonal argument. Finding a route into a difficult idea can require a story as well as a definition. His teaching makes room for both.

In August 2026, a public post introduced his conversation with Abhinav Singh at South Park Commons in San Francisco for Atoms to Intelligence. Singh was his former Masters student and a collaborator on the Hidden Markov Model research. Their subjects included chemical learning, molecular computing and stability as a possible limit on intelligence. A relationship formed around research had acquired a new setting and a new audience.

That exchange brings the story back to the doctoral office. Someone asks a question; another person helps make it worth pursuing. Years later, the roles change. Gopalkrishnan’s career has produced papers, courses and a company, but it has also produced conversations that continue after the original appointments end.

The computer he imagined at the start has gained unfamiliar surroundings: molecules, reactions, laboratory workflows. The questions have remained recognizable. How do parts organize? How does a system learn? What can be built from the answer? For Gopalkrishnan, those questions now have colleagues, customers and students attached. And somewhere in the personal record, a pair of bicycle tyres has already found employment as shoes.