Artificial intelligence has moved rapidly into software development, customer service and data analysis. Bringing the same technology onto a pharmaceutical or medical-device manufacturing floor is a different proposition.
In those environments, an impressive algorithm is only part of the equation.
Production systems must also account for traceability, validation, data integrity, quality controls and regulatory requirements. A technology that works successfully in a prototype may still have a long way to go before it can be trusted alongside systems responsible for manufacturing healthcare products.
That gap between what artificial intelligence can do and what regulated manufacturing can reliably use has become an area of focus for Satish Kumar Nalluri, a technology professional whose career has spanned Manufacturing Execution Systems, software engineering, automation and emerging AI technologies.
Nalluri has worked on manufacturing technology associated with companies including Corning, Johnson & Johnson Vision Care, Boston Scientific, Cepheid and Thermo Fisher Scientific.
The projects differ considerably in product and industry, but they share a common engineering challenge: turning increasingly complex physical production processes into reliable digital workflows.
As manufacturers now look toward AI, that experience is taking on a different relevance.
The Software Behind the Factory Floor
Manufacturing Execution Systems, or MES, occupy a largely invisible but important position in modern factories.
The systems sit between physical production operations and higher-level enterprise software, recording and controlling what happens as products move through manufacturing.
In regulated industries, they can also provide electronic production records, enforce process steps and maintain the traceability required to understand how a product was manufactured.
Nalluri’s earlier manufacturing work included MES modernization associated with Corning’s optical-fiber manufacturing operations.
His career subsequently shifted toward healthcare.
At Johnson & Johnson Vision Care, he worked with manufacturing systems supporting ACUVUE contact-lens production. At Boston Scientific, his experience included MES work associated with the WATCHMAN medical-device manufacturing environment.
The transition exposed him to a particular constraint of healthcare manufacturing: software changes cannot be evaluated solely by whether they make a process faster.
They also have to preserve control over how products are manufactured and documented.
That tension between innovation and control has followed Nalluri into his more recent work.
A Lesson From Molecular Diagnostics
At Cepheid, Nalluri worked on manufacturing-system enhancements associated with the company’s GeneXpert molecular-diagnostics platform.
His work included MES workflows, manufacturing-data and equipment integration, automation and improvements to software deployment processes.
Then came the COVID-19 pandemic.
Molecular diagnostics suddenly became part of a global effort to expand testing capacity. For engineers working behind those production systems, the crisis demonstrated how manufacturing software could become part of a much larger healthcare response.
It also illustrated a broader point about digital manufacturing.
When demand changes dramatically, the ability to scale production depends not only on physical equipment but also on the software coordinating manufacturing operations.
Nalluri subsequently joined Thermo Fisher Scientific, where his work has continued within life-sciences technology and digital manufacturing.
It is there that his longstanding MES background has increasingly intersected with automation and artificial intelligence.
AI Meets a Regulated Factory
Manufacturers have no shortage of potential AI applications.
Algorithms can search production data for unusual patterns. Predictive models can potentially identify emerging equipment or quality problems. Intelligent software can help teams interpret operational information, while robotic process automation can eliminate repetitive digital work.
But implementing those capabilities in regulated manufacturing is more difficult than demonstrating them in a laboratory.
A manufacturing environment may include MES platforms, equipment interfaces, enterprise applications, quality systems and years of established processes. New technologies have to operate within that ecosystem.
For Nalluri, the question is therefore less about whether AI can produce an impressive result and more about whether intelligent technologies can be incorporated into production systems without sacrificing reliability, traceability and compliance.
That is a considerably less visible part of the AI boom, but potentially an important one as artificial intelligence moves from experimentation into industrial infrastructure.
Taking Factory Problems Into Research
Nalluri’s interest in the subject does not end with corporate engineering projects.
Alongside his industry career, he has developed a scholarly research portfolio spanning artificial intelligence, automation, software systems and manufacturing technology.
His research has included work examining robotic process automation and digital workflow technologies in manufacturing, subjects that overlap with problems encountered in industrial environments.
His publications and their citations by other researchers are publicly documented through his Google Scholar profile, creating a public scholarly record extending beyond the companies where he has worked.
That distinction is relevant because much of industrial engineering remains invisible outside the organizations where it occurs.
Proprietary manufacturing systems are rarely described publicly in detail, even when the engineering challenges behind them are shared by companies across an industry.
Research provides another route for those ideas to circulate.
Through publication, technical approaches can enter the broader literature, where researchers and practitioners can examine them, challenge them and build upon them.
Nalluri’s career increasingly sits between those two environments: industrial implementation on one side and technical research on the other.
From Producing Research to Evaluating Technical Ideas
His engagement with the technical community has also expanded beyond authorship.
As his research activities developed, Nalluri became increasingly involved in the broader processes through which technical knowledge is evaluated and exchanged.
That dimension of engineering receives less attention than publication itself, but it is important to the way research communities operate.
New ideas in artificial intelligence, software engineering and automation are routinely tested through technical discussion and peer evaluation before they become part of the established literature.
For someone working in intelligent manufacturing, exposure to that broader research ecosystem can be particularly useful.
Industrial AI increasingly draws from several disciplines at once, including machine learning, software architecture, cloud computing, data engineering and automation.
The challenge is deciding which advances can survive the transition from research to production.
A Professional Network Spanning AI and Manufacturing
That multidisciplinary character is also visible in Nalluri’s professional involvement.
His affiliations have included organizations spanning engineering, pharmaceutical manufacturing, industrial automation and artificial intelligence, including the Institute of Electrical and Electronics Engineers (IEEE), the International Society for Pharmaceutical Engineering (ISPE), the International Society of Automation (ISA) and the Association for the Advancement of Artificial Intelligence (AAAI).
The combination reflects how difficult it has become to place intelligent manufacturing within a single technical discipline.
An AI application intended for a regulated factory may involve machine-learning models, manufacturing data, cloud infrastructure, automation systems, cybersecurity controls and domain-specific production knowledge.
In life sciences, quality and regulatory requirements add another layer.
The engineers working on those systems increasingly have to understand not only what new technologies can do, but also the environments into which they are being introduced.
Inside an Enterprise AI Lab
That convergence was visible in July 2025, when Nalluri was selected to support a U.S.-based Building Intelligent Apps laboratory involving Microsoft and NVIDIA technologies.
The hands-on program exposed Thermo Fisher participants to technologies associated with Microsoft Copilot, Azure OpenAI and NVIDIA’s AI ecosystem.
Thermo Fisher’s technology enablement leadership selected Nalluri as the company’s appointed contact for the U.S. laboratory, with responsibility for helping coordinate participant and vendor issues during the program.
It was one initiative within the much larger wave of enterprise AI experimentation taking place across corporate technology organizations.
But it also illustrates how companies are approaching the transition.
Rather than moving directly from a newly released AI model into production, organizations are creating controlled environments where employees can learn the technology, build applications and determine where it may actually be useful.
For manufacturers, that experimentation eventually encounters another hurdle: promising ideas have to survive contact with real production systems.
From Automation to Intelligent Manufacturing
Manufacturing automation itself is not new.
Factories have spent decades automating predefined processes. What artificial intelligence potentially changes is the ability of software to interpret larger amounts of information, recognize patterns and support decisions under changing conditions.
That could eventually affect how manufacturers approach quality monitoring, maintenance, production planning and process optimization.
Healthcare manufacturing raises the stakes.
A consumer AI application can make an inaccurate recommendation and inconvenience a user. A digital system participating in a regulated manufacturing process operates under a much higher expectation of reliability.
That is why some of the most consequential work in industrial AI may ultimately be less about developing another model and more about creating the architectures, controls and processes necessary to use intelligent systems responsibly.
Nalluri’s experience with MES platforms gives him a view of that problem from the factory floor upward.
His research provides another perspective, following developments in artificial intelligence and automation from the technical side.
The intersection is where much of his current work is taking place.
What Comes After the AI Experiment
The next phase of artificial intelligence in manufacturing may look less dramatic than the first.
The initial wave has been dominated by demonstrations of what generative AI and machine learning can do.
Industrial adoption will increasingly be about integration: connecting those capabilities to existing systems, determining where they create measurable value and establishing enough confidence in their behavior for operational use.
That transition is particularly relevant in biotechnology and life sciences, where manufacturers are simultaneously dealing with increasingly sophisticated products, growing volumes of data and demands for more flexible production.
The COVID-19 pandemic provided one illustration of why manufacturing agility matters. The continued development of genomics, personalized medicine and advanced therapies is providing another.
For engineers such as Nalluri, those developments are pushing traditional MES work into unfamiliar territory.
The manufacturing systems of the next decade may still perform their conventional functions of tracking materials, controlling workflows and documenting production.
But they may increasingly operate alongside intelligent systems capable of interpreting what is happening inside those workflows and helping people make decisions about what should happen next.
Making that combination useful, reliable and appropriate for regulated manufacturing could prove to be one of the less visible challenges of the artificial-intelligence era.
Unlike a chatbot demonstration, success will ultimately be measured on the factory floor.
