HELLO, I'MADEER KHAN

Adeer Khan

AI researcher passionate about developing smart solutions for the built world — applying software engineering, machine learning, and AI to create innovative solutions.

[ SEE MY RESUME ]
0+

Publications

0+

Projects

0

National Awards

WHAT I DO.

AI & ML researcher working on generative AI, LLM-integrated digital twins, computer vision, and intelligent built environments.

Development & AI Integration

Custom tools, workflows, and AI agents.

Connecting models, systems, and people across AEC workflows with practical AI integration.

  • -Training and evaluating deep learning models for vision, robustness, and generative tasks
  • -Designing neural network layers, including a network coding layer (NNCL) that recovers erased features under extreme erasure (IEEE Access, 2025)
  • -Applying ML to spatio-temporal and engineering data, e.g. ANN-CA urban growth modeling (EJRS, 2022)
  • -Generative AI for automated building design and AEC workflows across diffusion, autoregressive, and hybrid models
  • -LLM-integrated digital twin systems for management, adaptive control, and human-AI collaboration in architecture
  • -Building conversational agents with LangChain and OpenAI APIs
PythonPyTorchTensorFlowscikit-learnJupyterNumPyPandasHugging FaceLangChainOpenAIStreamlit

Urban Intelligence

From simulation to decision.

Turning spatial, visual, and environmental data into clearer decisions for intelligent built environments.

  • -Face recognition with deep embeddings (EE488B, KAIST)
  • -Point cloud rendering and architectural visualization in SketchUp and Enscape 3D
  • -Real-time digital twin communication with MQTT in Unity 3D
  • -Land use / land cover classification and urban growth modeling with random forests and ANN-CA
  • -Satellite imagery analysis and future-scenario simulation with QGIS and MOLUSCE
  • -Parametric environmental simulation with Rhino, Grasshopper, and Ladybug Tools
OpenCVUnityC#QGISRGit

Rapid Prototyping

Prototype fast. Customize at scale.

Moving research into working demos, reproducible systems, and production-ready experiments.

  • -Peer-reviewed publication in Automation in Construction (Q1), IEEE Access, and EJRS
  • -Structural health monitoring and experimental validation
  • -Docker, Linux, and reproducible research workflows
LaTeXDockerLinuxGitHub

EXPERIENCE.

NeoBIM GMBH logo

NeoBIM GMBH

AI Research & Development

Oct 2023 - Present

> Lead AI research and development for a BIM startup in Germany.

  • -Designing LLM-integrated digital twin systems and automated design workflows. https://designer.bluebrick.ai/
  • -Specializing generative models for the AEC industry, developing floorplan generation systems using Diffusion models, Autoencoders, Transformers, and 3D Gaussian Splatting (image/text-to-3D conversion).
  • -Evaluating 50+ generative AI repositories and fine-tuning models (Stable Diffusion, Flux, Mistral, Qwen, LLaMA) for scalable, domain-specific AEC design automation.
Korea Advanced Institute of Science & Technology (KAIST) logo

Korea Advanced Institute of Science & Technology (KAIST)

Graduate Researcher

Feb 2024 - Aug 2025

> Researched generative AI for automated building design and LLM-integrated digital twins.

  • -Implemented an LLM-assisted digital twin framework in Unity 3D, integrating IoT hardware (Arduino, ESP32) for intelligent building system simulation and control.
  • -Managed the lab server and website, and maintained the lab's IT infrastructure.
Joyn Group logo

Joyn Group

Data Scientist

Aug 2023 - Oct 2023

> Applied machine learning to business data and product workflows.

  • -Developed Urdu audio transcription pipelines with OpenAI Whisper, reducing word error rate from 70% to 22%.
  • -Trained DistilBERT for Named Entity Recognition in the cement industry, generating spatial market rate maps.
Softoo Pvt. Ltd. logo

Softoo Pvt. Ltd.

Asset Integrity Engineer

Apr 2023 - Aug 2023

> Worked with ML and computer vision models for structural inspection.

  • -Developed 3D point cloud generation from LIDAR for structure inspections.
  • -Improved redlining processes for offshore platforms, reducing time and costs.

EDUCATION.

Korea Advanced Institute of Science & Technology (KAIST) logo

Korea Advanced Institute of Science & Technology (KAIST)

MSc in Civil & Environmental Engineering

Feb 2024 - Aug 2025

Research on automated building design and LLM-integrated digital twins.

  • -CGPA: 3.50/4.30 (92/100%)
  • -Generative AI approaches for automated architectural design (Automation in Construction, 2025).
  • -LLM-integrated digital twin management and adaptive control for buildings.

PROJECTS.

[PROJ_001]

Digital Twin

A Unity3D-based digital twin system using MQTT for real-time bidirectional communication, visualization, and control between physical devices (Arduino/ESP32 sensors and actuators) and their virtual counterparts.

Face Embeddings

[PROJ_002]

Face Embeddings

A framework for training deep face-recognition embeddings - final project for Deep Learning for Computer Vision (EE488B) at KAIST, Fall 2024.

GITHUB
Podcast Summarizer

[PROJ_003]

Podcast Summarizer

OpenAI GPT-based podcast summarizer with a cloud-deployed backend and a Streamlit frontend. Paste any podcast RSS feed to get summaries and highlights.

LIVE DEMO
Talk with Images

[PROJ_004]

Talk with Images

A Streamlit app that answers questions about uploaded images through a LangChain and OpenAI GPT conversational agent.

GITHUB

[PROJ_005]

Al-Bahar Tower Facades

Designed the mechanical facades of Al-Bahar Tower in Rhino with parametric programming in Grasshopper, integrating Ladybug Tools to simulate facades opening and closing in response to the sun's path.

LIVE DEMO

[PROJ_006]

Game Development in Unity3D

A Unity3D project from the Computer Graphics Theory and Application (GCT522) course exploring procedural generation and shader-based weather effects in a dynamic environment.

Architectural Visualization

[PROJ_007]

Architectural Visualization

Building modeled in SketchUp and visualization render done in Enscape 3D.

LIVE DEMO
Point Cloud Rendering

[PROJ_008]

Point Cloud Rendering

Visualization of a point cloud model and a hillside tourist bridge in SketchUp, rendered in Enscape 3D.

LIVE DEMO

PUBLICATIONS.

[2025]

Generative AI approaches for architectural design automation

Automation in Construction

DOI
[2025]

Neural Network Coding Layer (NNCL): Enhancing Deep Learning Robustness against Feature Erasure

IEEE Access

DOI
[2022]

Machine learning-based monitoring and modeling for spatio-temporal urban growth of Islamabad

Egyptian Journal of Remote Sensing and Space Science

DOI
[2022]

A Simple and Sustainable Approach for Structural Health Monitoring of Structures

Capital University of Science and Technology

REACH OUT TO ME!

Discuss a project or just want to say hi? My inbox is open for all.

adeersafi@gmail.com
GITHUBLINKEDININSTAGRAM
adeerkhan avatar