Adaptive AI for the Field

Data & AI for
a Sustainable Earth

We find answers in data, and create value with AI

Deep AI earth is a Data & AI company spanning statistics and machine learning through spatial and vision AI. We analyze numerical, text, and satellite data to design models and AI for public-sector, real-estate, environmental, disaster, and healthcare problems. Drawing on years of research and consulting, we deliver analysis and AI tailored to each problem—not a single off-the-shelf solution.

Risk Detection View landslide risk-path service / simulation

Landslide-path risk analysis from satellite, terrain, and rainfall data

Very high
High
Moderate
Low

Why deep AI earth

Research-first

We start from original technology and research validated through papers, patents, and PoCs.

Adaptive AI

We build sustainable AI that adapts to changing data and environments—without retraining.

Cross-domain Analytics

We connect industries and problems through cross-domain analysis.

PoC-ready Partnership

We partner from rapid experiments and validation through to business scale-up.

UX Narrative Flow

Problem

The complex problems we set out to solve

Technology

Research-based technical hypotheses and approaches

Evidence

Research outcomes and objective evidence

Opportunity

Diverse domains and future opportunities

Company

People, vision, and the trust and capability of the company

Partnership

Partnership proposals and a call to act

Core Capabilities

Research-based, integrated AI capabilities

Big Data Consulting

From data strategy to architecture design and infrastructure

Machine Learning

Insights through statistical and ML modeling and feature engineering

Text Mining

Extract knowledge, sentiment, and topics from unstructured text

Deep Learning

Research and development of prediction, classification, and generative models

Data Fusion

Combining multi-source data with spatial, temporal, and semantic fusion

Predictive Analytics

Forecasting and scenario analysis based on time series, probability, and optimization

Application Domains

Arranged by demonstrated results

Each domain is labeled by validation stage, so the list reads as a hierarchy of evidence—not a catalog.

Proven

Forests · Wildfire · Landslides

We fuse satellite, weather, and terrain data to build risk maps and forecast likelihood.

Proven

Carbon · MRV

We quantify emissions, sinks, and reductions to support MRV and carbon management.

Proven

Environment · Climate

We analyze pollution, ecology, and water-resource risks to inform policy.

In progress

Cities · Disasters

We forecast compound hazards—flood, heat, earthquake—and analyze vulnerable areas.

In progress

Medical AI

We validate the robustness and reliability of imaging models to support reading assistance and early diagnosis.

In progress

Finance · Real Estate

We combine location, demand, and risk with spatial data to refine valuation.

Coming next · Defense AI / Agri AI / Climate tech / Humanities · Arts — designed together around how the client defines the problem.

CEO & Company

CEO Moon Hye-Jung

Moon Hye-Jung / http://moonhyejung.github.io/

FOUNDER & CEO

We start from research, answer the questions data has not yet explained, and design a better future with AI.

Grounded in data science and AI research, we solve complex problems across industries and build technology that creates social value.

Pursuit of knowledge

The power of data

Fairness in technology

Expansion of value

A sustainable future

Research is our starting point, data is our language,
and AI is the tool with which we serve the world.

Deep AI earth moves toward a wider field of possibility.

Technology · PoC · Partnership

Looking for a partner to solve your organization’s challenge with you?

Core Technology

Instead of rebuilding AI from scratch,
we adapt it to field data.

We implement AI that adapts quickly to field and domain data without large-scale retraining—and can explain and verify the grounds for its results.

01 Framework

Adaptive AI Engine

We keep the trained assets of existing models and apply them by correcting only for the characteristics of field data.

INPUT A

Existing AI · Open data

Use general-purpose models and open data as training assets

+
INPUT B

Field · Domain data

Incorporate field-collected data and equipment/environment conditions

→
CORE ENGINE

Lightweight adaptation engine

Lightweight adaptation and calibration via weight, statistics, boundary, and verification modules

RETRAINING-FREE
→
OUTPUT

Explainable, verifiable AI

Better performance + result explanation + automatic verification

02 Modules

Six core modules

01

Weight Optimization

Weight Optimization

We lightly adjust model weights to the field-data distribution to optimize performance.

02

Statistical Calibration

Statistical Calibration

We calibrate class, pixel, and ratio statistics to the field distribution to reduce prediction bias.

03

Multi-region Detection

Multi-region Detection

We split large areas into regions to raise both detection performance and efficiency.

04

Unknown Detection

Unknown Detection

We detect objects and patterns the model was never trained on, and surface risk or anomaly signals.

05

Boundary Auto-detection

Boundary Auto-detection

We automatically detect object and region boundaries and keep improving precision.

06

Auto Ground-Truth Verification

Auto Ground-Truth Verification

We automatically verify predictions against multiple rules and evidence, and quantify reliability.

03 Stack

Integrated AI technology stack

Big Data Consulting

  • Data strategy · governance
  • Collection · quality · infrastructure

Machine Learning

  • Statistical modeling
  • Feature extraction · selection

Text Mining

  • Natural language understanding (NLP)
  • Document classification · summarization

Deep Learning

  • CNN · RNN · Transformer
  • Image · video · time series

Data Fusion

  • Spatial · temporal fusion analysis
  • Multi-source data linkage

Image · video AI

  • Object detection · classification · tracking
  • Satellite · aerial · drone

Predictive Analytics

  • Demand · risk · performance forecasting
  • Scenario analysis

Explainable AI

  • Visualization of prediction evidence
  • Reliability · regulatory response

Technical review · joint research · PoC · investment talks

We design an AI strategy that fits the actual problem in the field.

Application Domains

We do not sell a fixed solution.
We define the problem together.

We understand the real problem in the domain, form a hypothesis with data and AI, and validate it through a PoC.

Proven

Proven

Proven

Forests · Wildfire · Landslides

We fuse satellite, weather, and terrain data to continuously monitor wildfire risk and landslide likelihood.

Proven

Carbon · MRV

We quantify carbon emissions, sinks, and reductions to support MRV and carbon management.

Proven

Environment · Climate

We analyze pollution, ecology, and water-resource risks in data and propose ways to manage them.

In progress

In progress

Cities · Disasters

We forecast compound hazards—flood, heat, earthquake—and reduce damage through vulnerability analysis.

In progress

Medical AI

We validate the robustness and calibration of imaging models to support reading assistance and early diagnosis.

In progress

Finance · Real Estate

We combine location, demand, price, and risk with spatial data to refine valuation.

Future AI Opportunities

Defense AI

Risk prediction and detection, plus support for tactical decisions

Agri AI

Precision agriculture, yield forecasting, and productivity gains

Climate tech

Discovering solutions for climate risk, adaptation, and mitigation

Carbon map

Maps of green, blue, gray, and dark carbon

Humanities · Arts

Analysis of cultural-content and heritage data

What we do

What we can do

Data Fusion

We unify satellite, IoT, sensor, administrative, and open data into a single analytic system.

Data analysis

We surface hidden patterns and insights through statistical, spatial, and time-series analysis.

Predictive modeling

We quantify future change and risk with machine learning and deep learning.

Decision support

We present evidence and alternatives together through dashboards, maps, and visualization.

We look for new opportunity in your problem.

Let’s create value together through research, analysis, and PoC collaboration.

Company

A research-based AI startup

Deep AI earth is a research-based AI startup that repeatedly applies adaptive AI—validated on satellite and spatial data—across changing domains. Instead of selling one solution, we stand up the right AI for each problem, quickly.

CompanyDeep AI earth
CEOMoon Hye-Jung — Founder & CEO / http://moonhyejung.github.io/
FoundedJuly 14, 2026
Office10F, 130 Wangsimni-ro, Seongdong-gu, Seoul (Seongsu-dong 1-ga, KCC Premier Tower)
Phone02-738-3486
Core valuesTrust · Innovation · Collaboration · Impact
CEO Moon Hye-Jung

Moon Hye-Jung / Founder & CEO

FOUNDER & CEO

We start from research, answer the questions data has not yet explained, and design a better future with AI.

AI research and innovation

We develop distinctive solutions through research on deep learning and original AI technology.

Big-data analysis

We analyze large-scale data with precision and turn it into insight.

Cross-domain research

We connect industries and domains to propose new ways of solving problems.

Field-first problem solving

We do not stop at research—we turn it into solutions that can be applied.

Mission & Vision

Trustworthy AI

Trustworthy AI

Trust in the data, the model, and the result comes first.

Adaptive AI

Adaptive AI

We build AI that adapts flexibly to environments and change.

Research to Impact

Research to Impact

We turn research outcomes into real-world impact.

Cross-domain Intelligence

Cross-domain Intelligence

We combine domain knowledge with AI to create fused value.

Sustainable Value

Sustainable Value

We create value that has a positive effect on society and the environment.

Roadmap

Growth roadmap

PHASE 01

Research

We develop core technology through foundational research and data acquisition.

→
PHASE 02

PoC

We apply AI to field problems and run a verifiable PoC.

→
PHASE 03

Partnership

We expand partnerships and build a collaboration ecosystem.

→
PHASE 04

Scale

We realize sustainable growth through commercialization and scale.

Research is our starting point, data is our language,
and AI is the tool with which we serve the world.

Deep AI earth moves toward a wider field of possibility.

Contact & Partnership

Tell us the problem and the data.
We start by proposing a method that can be validated.

Technical review, joint research, a PoC, or an investment discussion—any starting point is fine.

Technical review

We review technical feasibility and how it could be applied to your organization’s problem.

Joint research

We co-develop new technology and solutions with research institutions and companies.

PoC in the field

We quickly validate whether the technology works—and whether it has value—on real data.

Investment talks

We review strategic investment in projects with strong potential to grow.

CONTACT

02-738-3486
contact@deepaiearth.com
www.deepaiearth.com

Directions

Deep AI earth, 10F, 130 Wangsimni-ro, Seongdong-gu, Seoul(Seongsu-dong 1-ga, KCC Premier Tower)
Deep AI earth location map — Seongsu-dong KCC Premier Tower

200 m walk from Ttukseom Station, Exit 1