生命科學學生以 AI 探索藥物發現與醫療保健，並持續開發項目

I'm a Life Sciences Student Building With AI, and I'm Just Getting Started

Most people studying my degree spend their time learning about cells, genetics, biochemistry, and living organisms.

I do that, too.

But somewhere along the way, I became curious about something else: What happens when you combine biology with artificial intelligence?

Could machine learning help us understand biological data? Could computational methods help researchers explore potential drug candidates? Could AI make healthcare tools more accessible?

I didn't have all the answers.

So instead of waiting until I knew everything, I started building.

From learning to experimenting

My journey into technology wasn't a straight line.

I started exploring data analytics, learning tools like Python, SQL, Excel, and Power BI. Eventually, my curiosity expanded into machine learning, bioinformatics, and computational drug discovery.

At first, there were plenty of tutorials and notes.

But I realized something: understanding a concept and using it to solve a problem are two very different things.

I needed projects that would force me to apply what I was learning.

That's when I started experimenting with ideas that connected my academic background with technology.

𝐏𝐫𝐨𝐣𝐞𝐜𝐭 1: 𝐄𝐱𝐩𝐥𝐨𝐫𝐢𝐧𝐠 𝐦𝐚𝐜𝐡𝐢𝐧𝐞 𝐥𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐟𝐨𝐫 𝐝𝐫𝐮𝐠 𝐝𝐢𝐬𝐜𝐨𝐯𝐞𝐫𝐲

One of the areas that fascinates me most is computational drug discovery.

I began exploring molecular docking workflows and working on a project called DockLens, which combines machine learning with molecular docking analysis.

The idea was to explore how machine learning could help estimate molecular binding affinity and complement traditional docking workflows.

Working on this project introduced me to tools such as AutoDock Vina, Python, and molecular visualization software.

It also taught me an important lesson: a model producing a prediction doesn't automatically mean the prediction is scientifically reliable.

You need to understand the data, evaluate the model, examine its limitations, and interpret the results carefully.

I'm still developing this project, but the process has made computational biology feel much more tangible than simply reading about it.

𝐏𝐫𝐨𝐣𝐞𝐜𝐭 2: 𝐄𝐱𝐩𝐥𝐨𝐫𝐢𝐧𝐠 𝐀𝐈 𝐟𝐨𝐫 𝐡𝐞𝐚𝐥𝐭𝐡𝐜𝐚𝐫𝐞

Another project I'm working on explores a Type 2 Diabetes Digital Twin focused on glucose spike prediction.

The idea is to investigate whether patient characteristics and time-series data can help build models that predict changes in glucose levels.

For the initial development, I've been working with synthetic patient data and simulated wearable measurements.

My workflow involves preparing datasets, organizing patient-level information, and exploring how machine learning could be applied to glucose prediction.

One important distinction: synthetic data and model predictions are useful for experimentation, but they don't establish clinical validity.

There's still a lot to learn about feature engineering, evaluation, temporal patterns, and the limitations of predictive models in healthcare.

That's precisely why I find the project interesting.

It gives me a way to explore machine learning while remaining connected to my background in life sciences.

Where AI fits into my workflow

I also use AI tools to help me learn and build.

When I encounter an unfamiliar concept, I use AI to explore explanations. When I run into an error, I use it to investigate possible causes. When I'm planning a feature, I use it to compare approaches and break the work into smaller tasks.

But I don't want my entire workflow to become copying code that I don't understand.

My goal is to understand what I'm building, test whether it works, and learn from the mistakes along the way.

AI can accelerate parts of the process, but it doesn't remove the need for critical thinking.

If a model produces an unexpected result, I still need to investigate it. If a program runs successfully, I still need to determine whether it does the right thing.

For me, AI is most useful when it helps me become a better problem-solver, not just a faster code generator.

The part that doesn't fit into a portfolio

I'm also a full-time undergraduate student.

Between college classes, practicals, assignments, examinations, and a long daily commute, finding time for independent learning can be difficult.

I can't always spend an entire day working on a project.

Sometimes progress means fixing one bug. Sometimes it's understanding one algorithm. Sometimes it's spending an hour figuring out why a workflow failed.

I've learned to appreciate those small steps.

I'm not trying to pretend that the journey is easy or that I've mastered every technology I use.

I'm learning, experimenting, and trying to build a stronger foundation with the time and resources available to me.

What's next?

I'm interested in exploring the intersection of:

• Artificial intelligence and machine learning
• Bioinformatics and computational biology
• Computational drug discovery
• Healthcare analytics and predictive modelling
• Data science and engineering

I want to develop projects that go beyond demonstrations and gradually become more reliable, reproducible, and useful.

I also want to document what I learn so that other students who are starting from a similar place can see both the progress and the problems behind the finished result.

I'm still early in this journey.

But I don't want to spend years only preparing to build something.

I'd rather keep learning, keep building, and improve through the process.

If you're working in AI, bioinformatics, drug discovery, or healthcare technology, I'd love to hear what you're building and what you've learned along the way.

You can explore my projects and follow my progress here:

GitHub: https://github.com/SanchitKRai

Thanks for reading!

I'm a Life Sciences student exploring the intersection of biology, machine learning, and AI. I'm documenting my journey one project at a time.