Build Scientific Apps Without Traditional Coding

Empowering scientists to rapidly explore, analyze, and share data

Transform from prompting ChatGPT to building effective data science applications using AI-assisted workflows. Go from idea to working application in hours, not weeks. (Production deployment requires additional AI Engineering skills)

Start Learning →

Who Am I?

Dr Raminderpal Singh - Engineer at heart with a focus on applying engineering thinking to drug discovery analytics. Especially interested in the intersection of drug discovery data, LLM processing, and real-world workflows.

→ View LinkedIn Profile

Current Focus

Write LLM-focused apps for drug discovery
Apps for data scientists & wet lab scientists
Bridging computational and experimental workflows

Leadership Roles

Global Head of AI & GenAI Practice, 20/15 Visioneers
Founder, HitchhikersAI grassroots community

https://www.20visioneers15.com/

Why Vibe Coding?

Vibe coding takes everyday prompting with ChatGPT/Claude and adds structure and workflow to create effective data science applications. Rapidly explore, analyze, and share data at minimal cost. (Production deployment with scalability, security, and enterprise features requires additional AI Engineering skills)

This vibe coding method focuses on this intersection

Vibe coding: intersection of domain knowledge, scientific questions, and LLM processing

From Everyday Prompting to Data Science Apps

⚡

Rapid Development

Go from concept to working data science application in hours. Build tools for data exploration, analysis pipelines, and interactive dashboards for research and discovery.

💰

Low Cost

Free or low-cost tools (Claude, Cursor AI, Vercel). No expensive enterprise software. Perfect for academic labs and small biotech teams with limited budgets.

🔬

Scientist-Friendly

Focus on scientific logic, not programming syntax. Describe what you want in plain English. AI handles the implementation details and code generation.

Example: Cell Confluence Analyzer

A real data science application built using vibe coding. Upload cell microscopy images and get confluence percentage analysis with visual overlays. Built in ~2 hours using the workflow below. Runs locally at http://localhost:8000 (optional deployment guide included).

Simple English Workflow

Describe to Claude

"Create an app that measures cell confluence from microscopy images using established Thermo Fisher methodology."

Claude Designs

Claude provides UI mockups, data flow diagrams, and architecture recommendations. No code yet—just design artifacts.

Cursor Implements

Give Claude's designs to Cursor AI. It generates the complete FastAPI application with image processing logic.

Test & Iterate

Run locally at localhost:8000, test with real images, iterate on features. Optional: Deploy to Vercel to share with colleagues.

→ See detailed walkthrough in the guide

Application Output

Upload image → Get confluence percentage + visual overlay

Example Input Images

Cell microscopy images at different confluence levels

Upload Interface

Analysis Results

What You Get

✅

FastAPI data science app

✅

Image processing pipeline

✅

Local web interface

Optional: Deployment guide included for sharing with colleagues

Watch It Being Built

See the complete build process for the Confluence Assessment demo. This unedited recording shows the actual workflow—Claude generates designs, Cursor writes code, bugs get fixed, features get refined. Watch a working data science application come together from scratch.

▶ Watch the Build Process

Ready to Start Building?

Follow the complete guide to build your first vibe-coded data science application. No prior programming experience required.

Access the Full Guide →