About

Hello! I'm a Data Scientist and AI Researcher based in Vancouver, Canada. I specialize in deep learning, computer vision, and building human-in-the-loop AI systems that solve real-world problems.

With a background in generative models and semantic segmentation, I've published research at top venues like CVPR and built production AI systems for autonomous inspection and satellite imagery analysis. I'm passionate about bridging the gap between cutting-edge research and practical applications.

Projects

Human-in-the-Loop AI System for Autonomous Inspection

End-to-end AI-assisted system for automated inspection with secure Azure integration and real-time inference pipeline. Built React-based annotation interface reducing annotation time through optimized canvas rendering.

PyTorch Azure React Computer Vision

Continuous Learning Pipeline for AI Models

Automated pipeline for processing log data to continuously update and fine-tune AI models. Improved model performance through regular retraining, ensuring adaptability to changing conditions.

MLOps CI/CD Deep Learning

Land Cover Segmentation System

AI framework for segmenting satellite images and data imputation systems for near-detection-limit scenarios in geospatial analysis.

Semantic Segmentation Satellite Imagery TensorFlow

GAN Debugging Framework

Novel post-hoc debugging framework for improving quality of trained generative vision models without retraining. Published at CVPR 2021.

GANs PyTorch Generative Models

Skills

Machine Learning and AI

Python PyTorch TensorFlow Computer Vision Generative Models Vision-Language Models

Specializations

Semantic Segmentation Human-in-the-Loop Systems Reinforcement Learning GANs

Engineering and Infrastructure

Azure AWS React CI/CD Git

Resume

Job Experience

Data Scientist

Sep 2022 - Present

ALS Geoanalytics, North Vancouver, BC

Developed end-to-end AI-assisted systems for automated inspection with Azure integration. Built React-based annotation interfaces and continuous learning pipelines for model updates. Created AI frameworks for satellite image segmentation and data imputation.

AI Researcher

Oct 2019 - June 2022

KAIST, Graduate School of AI

Developed novel post-hoc debugging framework for generative vision models (CVPR 2021). Research covered by UNITE AI, KAIST Breakthroughs magazine, and Korea AI-Times.

Education

MSc. Computer Science and Engineering

Aug 2016 - Aug 2018

UNIST (Ulsan National Institute of Science and Technology)

Thesis: Deep fully residual convolutional neural network for semantic image segmentation

BSc. Computer Science

Aug 2011 - Jun 2016

Sharif University of Technology

Download Resume

Publications

Automatic Correction of Internal Units in Generative Neural Networks

CVPR 2021

A. Tousi, H. Jeong, et al. Novel framework for debugging and improving quality of trained GANs without retraining.

Featured in UNITE AI, KAIST Breakthroughs magazine, and Korea AI-Times

Deep Fully Residual Convolutional Neural Network for Semantic Image Segmentation

Master's Thesis, UNIST 2018

Research on advanced architectures for semantic segmentation using fully residual connections.

Blog

Dec 2024

Understanding Transformer Architectures

A deep dive into attention mechanisms and how transformers revolutionized NLP...

Nov 2024

Optimizing PyTorch Training Pipelines

Tips and tricks for speeding up your deep learning training workflows...

Oct 2024

The Future of Multimodal AI

Exploring the convergence of vision and language models...

Contact

I'm always open to discussing new projects, research collaborations, or opportunities in machine learning and AI.

LinkedIn linkedin.com/in/litoali/
GitHub github.com/alitou