How Vijay Kumar Naidu Velagala Is Building the AI Driving Smarter Pharma Research

In order to develop safe and effective medical treatments, pharmaceutical companies generate massive amounts of data across experiments, clinical trials, and exploratory research.

However, properly leveraging this data remains a challenge. Many organizations still rely on manual workflows that slow progress and introduce human error, while even modern analysis tools still struggle to handle the complexity of increasingly growing datasets. As a result, researchers lose time on repetitive tasks, overlook key insights, and face costly delays in developing life-saving treatments.

Vijay Kumar Naidu Velagala has been working to address this gap as a data scientist, spending the past two years building custom AI systems to improve workflows across pharmaceutical organizations.

His tools aim to allow research teams to automate routine work, analyze complex datasets more effectively, and make faster, more informed decisions earlier in the drug development process—helping vital medicines reach patients with greater confidence.

Improving Molecule Screening in Early Drug Development

Having graduated with a master's in chemical engineering, Vijay has spent his career working as a data scientist in fields like biology and biochemistry. Since 2023, he's been working to build and deploy AI platforms that help pharmaceutical companies interpret complex scientific data and speed up drug development.

One major project Vijay led focused on improving the early selection and testing of molecules, an essential first step in identifying potential drug candidates before they enter preclinical studies.

This is a stage where many research teams struggle, often selecting compounds with poor properties like low solubility, poor absorption, or rapid metabolism. This leads to high failure rates, with over 90% of tested candidates failing to progress past Phase I clinical trials—delaying treatment development for serious diseases.

To help one client address this challenge, Vijay built a system powered by graph neural networks, a type of AI model designed to understand how individual components interact within a connected structure. In this case, the model learned how different relationships between atoms and bonds, the main components of a molecule, influence its behavior in the body.

This allowed researchers to better predict factors in molecular structures like absorption, distribution, metabolism, excretion, and toxicity (ADMET), which are crucial to determine whether a compound is likely to succeed as a drug. With these insights, teams could screen molecules more effectively, eliminate weak candidates sooner, and boost the chances of success in later-stage trials.

Speeding Up Antibody Design Protocols

Building on this work, Vijay also led a project focused on antibody design, a more specialized area of drug development focused on creating proteins that can identify and neutralize harmful agents in the body (such as viruses or bacteria).

Traditionally, scientists would manually test numerous biological sequences to find those that perform well under specific conditions, an approach that can be time-consuming and risks producing proteins that fail in real-world settings.

To make this process more efficient, Vijay trained a large-scale AI model on datasets of protein sequences, teaching it how different patterns influence biological behavior. In doing so, his platform could generate optimized protein candidates tailored to specific therapeutic goals, giving researchers high-probability options without the need to manually sift through vast libraries of designs—speeding up early-stage development and giving drug discovery teams the tools to bring promising candidates into preclinical testing more quickly.

Giving Researchers Better Access to Critical Data

Vijay has also leveraged AI to help scientists more easily access and apply the research behind those discoveries. This remains a major challenge in the field, as valuable insights are often buried across disconnected sources like clinical notes, lab reports, and prior studies—forcing researchers to spend hours piecing things together before they can move forward.

One of Vijay's clients had tried to solve this with a conversational AI trained on internal research data. But the model struggled with complex queries that required broader context, often returning incomplete or misleading information.

To fix this, Vijay rebuilt the system using a retrieval-augmented generation framework. This way, instead of relying only on its training data, the assistant could pull relevant information from outside sources, resulting in answers that were more tailored to specific needs and, according to Vijay, up to 30% more accurate.

Helping Manufacturers Ensure Drug Safety

Over the years, Vijay has also worked on several manufacturing projects, assisting pharmaceutical companies in improving their production workflows and ensuring the medicines they release remain safe and effective throughout their shelf life.

In one initiative, Vijay developed a system that analyzes how various manufacturing variables (such as mixing speed, cooling rates, or raw material differences) can impact a drug's long-term stability. By helping manufacturers pinpoint which process changes result in more consistent batches, this tool gives them the insight to adjust their methods early and reduce the risk of stability issues after the product has already shipped.

In a separate project, he designed an application that models how a drug chemically degrades under different storage conditions. The model uses reaction kinetics (mathematical models that describe how chemical compounds respond to factors like heat, humidity, or light) to estimate how quickly a drug might lose potency, enabling manufacturers to anticipate storage risks and make better decisions about packaging, environmental controls, and expiration dates.

The goal of these projects is simple: reduce the risk of product failure, meet regulatory expectations, and deliver medicines with greater confidence in stability and performance.

Streamlining Scientific Research with AI

By developing advanced AI platforms that evaluate promising drug candidates, design more optimal antibody sequences, and organize complex research data, Vijay Kumar Naidu Velagala is applying artificial intelligence to some of the most critical workflows in pharmaceutical development—aiming to give scientists the clarity they need to develop treatments more efficiently and effectively.