Aryan Jainaryanj {at} mit {dot} edu

Kisho: The AI Data Scientist

9/3/2024

Note: Post is from YC's Launch YC forum. Represents idea #5 during the batch.

In the rapidly evolving landscape of data science and machine learning, a new tool is emerging to democratize access to advanced analytics. Kisho, developed by a talented team from MIT, is revolutionizing how we approach data science through AI-powered automation.

The Challenge

Domain Experts Struggle to Leverage Their Data

Professionals across various sectors - healthcare, finance, and scientific research - possess invaluable domain knowledge and data. However, they often lack the technical expertise required to apply sophisticated analytical techniques. This gap results in missed opportunities for innovation and data-driven decision-making.

Data Scientists Face Workflow Bottlenecks

Even seasoned data scientists find themselves spending considerable time on repetitive tasks:

  • Data cleaning
  • Feature engineering
  • Model selection

This routine work reduces the time available for high-value activities such as result interpretation and strategic planning.

The Solution

Kisho serves as your AI-powered data science assistant, seamlessly integrated into Jupyter Notebooks. The process is remarkably simple:

  1. Upload your data
  2. Describe your problem or goal in plain English
  3. Watch as Kisho automatically writes, executes, and refines code to accomplish your task

Key Features

Natural Language Interface

Communicate your data analysis needs in plain English. Kisho translates your intentions into executable code.

Automated Workflows

Kisho manages the entire data science pipeline automatically, from data cleaning to model training and evaluation.

Self-Healing Code

The AI system detects and fixes errors in real-time, eliminating frustrating debug sessions. This capability extends to overnight training runs, saving valuable time and compute resources.

The Team

👋 Meet Aryan and Ayush, the minds behind Kisho. Their journey began during their first year at MIT, followed by extensive experience in machine learning research and countless hours in Jupyter Notebooks.

Aryan's Background

  • Developed computer vision models at Al Gore's Climate TRACE, with work published at AAAI
  • Scaled a climate-compliance app to 45 schools nationwide
  • Contributed to MIT CSAIL's research on cancer cell detection using foundational models

Ayush's Experience

  • Collaborated on exoplanet discovery with JPL
  • Worked on satellite systems at Northrop
  • Pioneered development of a new 3-valued quantum dot transistor

Get Involved

The team is actively seeking early adopters for both enterprise and research applications. If you have valuable data but lack technical resources for analysis, you can:

  • Join the waitlist
  • Contact the founders directly at founders@kisho.app

Let Kisho supercharge your data science workflows and unlock the full potential of your data.