CV
Download CV
Detailed CV
Education
- PhD in Computer Science (EU’s Marie Skłodowska-Curie Fellowship) - 2020 - 2024
University Côte d’Azur, Nice, France
BiomechanicsCapture: 3D reconstruction of the human body using deep learning and neural rendering- Led pioneering project on real-time human biomechanics capture using deep learning and neural rendering techniques (NeRF, Gaussian Splatting)
- Published research findings in international conferences, contributing to the scientific community
- Industry partner: Youdome, Monaco
- Academic partners: Brown University, United States
- MSc in Computer Science & Engineering (Specialized in Artificial Intelligence) - 2017 - 2020
Politecnico di Milano, Milan, Italy- Mastered AI, Machine Learning, Data Science, and Computer Vision through intensive, project-based coursework
- Developed robust analytical foundation in mathematics, probability, statistics, and critical thinking
- Key projects: “Art with computer vision”, “Rob the Dancing Robot”, “Robotics with ROS and Gazebo”
- B.Tech in Computer Science & Engineering - 08/2013 - 08/2017
West Bengal University of Technology, Kolkata, India- Final thesis on automated region selection in images based on visual saliency ____________
Work experience
- Senior Machine Learning Engineer - 2025 - Present
Haxion AI, San Francisco, United States - Remote from Spain- Designed and deployed production-grade LLM pipelines, enabling real-time natural language interaction with complex 3D datasets
- Developed an iOS mobile app using ARKit to capture 3D scene using Lidar and cameras
- Developing a pipeline to generate 3D scenes along with their corresponding 3D features from videos
- Delivered end-to-end prototypes to client-ready solutions, ensuring model validation and deployment feasibility
- Tools: Python, GenAI, LLMs finetuning, VLMs, Gaussian Splatting, UV, Pydantic, FastAPI, AWS, and iOS
- Visiting Researcher - 2022 - 2024
Brown University, Providence, USA- Key Contribution: Designed and executed algorithms for 3D human digitization using Neural Rendering techniques (e.g., NeRF, Differentiable Rendering)
- Optimization: Improved training performance for large-scale neural rendering models by implementing multi-GPU distributed AI pipelines
- Research Impact: Published peer-reviewed papers in CVPR conferences, contributing to the wider research community
- Teaching Assistant - 2022 - 2023
University Côte d’Azur, Nice, France- Taught Computer Vision and Machine Learning courses to Master’s students
- Machine Learning Engineer - 2020 - 2020
Next Industries S.R.L, Milan, Italy- Developed and deployed machine learning models for wearable edge devices with a focus on motion gesture recognition in real-time
- Collaborated with cross-functional teams including hardware and software engineers to align ML algorithms with business goals
- Data Scientist Intern - 2019 - 2020
UniCredit Services S.c.p.a., Milan, Italy- Automated and optimized financial processes using NLP and deep learning, resulting in 40% reduction in manual processing time
- Deployed NLP microservices in distributed architecture aligned with data governance principles
- Collaborated with product teams to deliver AI-driven features that improved user experience
- Tools: Python, Spacy, OpenCV, GitLab, MLflow, scikit-learn, Docker, Kubernetes
Skills
Core Competencies: Generative AI, LLMs, NLP, CV, Machine Learning, Deep Learning
Programming: Python (Extensive), C, C++, Java, Matlab, and SQL
Generative Models: LLMs, VLM, GenAI, Diffusion Models, GANs, VAEs, Hugging Face Transformers, and RAG
AI Research & Experimentation: Model Training, Finetuning (LoRA/QLoRA/PEFT), Benchmarking, Prototyping
ML & CV Tools: PyTorch, TensorFlow, PyTorch Lightning, FastAPI, Scikit-learn, Pandas, OpenCV, CUDA GPU
DevOps: Docker, CI/CD, MLflow, GitHub, GCP (Kubernetes Engine, Compute Engine), AWS (EC2), Amazon Bedrock
Languages: English (Proficient), Hindi & Bengali (Native), Italian (Beginner), Spanish (Beginner)
Publications
Arnab Dey, Andrew I. Comport. RGB-D Neural Radiance Fields: Local Sampling for Faster Training
Dey, A. and Ahmine, Y. and Comport, A.I., Mip-NeRF RGB-D: Depth-Assisted Fast Neural Radiance Fields, Journal of WSCG, Volume 30 pages 34-43, DOI:10.24132/JWSCG.2022.5.
PNeRF: Probabilistic Neural Scene Representations for Uncertain 3D Visual Mapping. Y Ahmine, A Dey, AI Comport - arXiv preprint arXiv:2209.11677, 2022
Dey, Arnab, et al. "GHNeRF: Learning Generalizable Human Features with Efficient Neural Radiance Fields." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024.
Dey, Arnab, et al., "HFNeRF: Learning Human Biomechanic Features with Neural Radiance Fields," 2024 IEEE Conference on Artificial Intelligence (CAI), Singapore, Singapore, 2024.
Lu, Cheng-You, et al. "DiVa-360: The Dynamic Visual Dataset for Immersive Neural Fields." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024.
Arnab Dey, Cheng-You Lu, Andrew I. Comport, Srinath Sridhar, and Jean Martinet. "HFGaussian: Human gaussian with biomechanics features". arXiv preprint arXiv:2411.03086 (2024).
Thesis
- A. Dey, Human activity and emotion recognition from RGB videos using deep learning. Politecnico di Milano, a Deep learning-based approach to recognize human activities and emotions from real-time RGB videos for robotics applications, Implemented using PyTorch.
- A. Dey, Automated region selection in images based on visual saliency. West Bengal University of Technology, Achieved automated region selection without prior knowledge and user interaction, Implemented using MatLab.
Teaching
Awards and Certificates
- 2022 Best poster award for “RGB-D NeRF: Local Sampling for Faster Training,” EuroGraphics 2022, Reims
- 2022 Winner of E-Health Creathon Innovation Center for Entrepreneurship of University Côte d’Azur
- 2020 EU’s H2020 Marie Skłodowska-Curie Fellowship
- 2019 Developed official IOI mobile application, The International Olympiad in Informatics