Obscura - No-Code AI Analytic Platfrom

"Upload your data. Get analysis, insights, and visualizations instantly — without writing code."

Problem – Non-technical researchers struggle to analyze data independently

Graduate students, business analysts, and field researchers frequently rely on others to perform even simple analyses due to:

• Limited coding skills (Python/R)

• Difficulty creating professional visualizations

• Time-consuming back-and-forth when requesting help

• Lack of automated tools that understand both data and analytical intent

As a result, valuable insights are delayed, inconsistent, or never fully explored.

Insight – The barrier is not data itself, but the need to translate ideas into code

In interviews with early users, the common pain point was:

“I know what I want to analyze. I just don’t know how to write the code.”

Most users understood their data and research questions, but lacked the ability to:

• Select variables

• Apply the right analytical method

• Generate compelling charts

• Summarize insights clearly

This revealed a gap: data interpretation tools must remove the need for code, not the need for thinking.

Solution – Obscura

An AI-native platform for instant data exploration, analysis, and visualization — no coding required.

Key Features

1) Automatic Data Profiling

• Upload CSV/Excel

• Schema, data types, distributions, missing values analyzed instantly

• Automatic detection of continuous/categorical variables

2) Built-in Basic Analysis (V0.0.1 scope)

• Summary statistics

• Correlation checks

• Outlier detection

• Frequency tables

• Automated EDA report generation

3) AI-Guided Natural Language Analysis

• Users can ask: “Show me the relationship between age and spending.”

• The agent selects appropriate variables

• Generates correct plots (scatter, bar, boxplot, heatmap)

• Provides narrative explanations

4) One-click Visualization Export

• High-quality plots for reports and papers

• Clear color schemes optimized for academic/industry presentations

User Flow

Upload File
→ Automatic Profiling
→ (Optional) Natural Language Request
→ Variable Selection & Chart Generation
→ AI Interpretation
→ Exportable Report

My Role – Founder, Product Designer & Builder

I led Obscura from concept to functional prototype with real beta testers.

1) Problem Discovery & Product Positioning

• Conducted interviews with graduate students and early-career researchers

• Identified the core gap: need for insights without coding

• Positioned Obscura as a no-code alternative to manual Python analysis

2) UX & Product Architecture

• Designed the 5-step product structure:

  1. API setup (internal)

  2. Data exploration

  3. Preprocessing setup

  4. Feature list configuration

  5. Natural language analysis

• Built the initial V0.0.1 prototype focusing on automatic data exploration

3) AI Workflow Design

• Developed intent classification for analysis requests

• Designed variable selection heuristics

• Integrated chart-type selection logic based on data characteristics

4) Engineering Implementation

• Built backend using Python + FastAPI

• Managed file parsing, data validation, preprocessing

• Implemented analysis and visualization engine

• Integrated LLM for interpretation and user guidance

5) Beta Tester Program (40 Users)

• Launched early user cohort from university communities

• Collected structured feedback

• Improved UX, error handling, and chart quality based on user sessions

Impact – A practical research tool for non-technical users

• Enabled users to perform EDA without writing any code

• Delivered instant charts comparable to Python/Matplotlib quality

• Automated entire steps: profiling → visualization → interpretation

• Validated user need through 40 beta testers

• Provided a clear foundation for next versions (V0.1 modeling stage)