Surgical AI · VirtaMed
AI Proctor
A real-time AI coach that watches surgery training and talks you through it.
About
Research curiosity, with an engineer’s habit of shipping.
For the past 4.5 years, I’ve been a Machine Learning Research Engineer at VirtaMed in Zurich, building AI for surgical training. I take projects from model development to browser-based applications and deployment, including computer vision, generative image enhancement and AI coaching systems.
At ETH Zurich, I completed a master’s in Computational Science and Engineering, focusing on robotics. I co-authored a WACV paper on neural 3D reconstruction, co-invented a camera pose estimation method filed as an international patent application, and worked on de novo drug design in the Molecular Design Lab.
I graduated with honours from the Moscow Institute of Physics and Technology. During an Amgen Scholars internship at ETH Zurich, I used generative adversarial networks to study Alzheimer’s-related brain changes and presented the work at the Amgen Scholars Symposium in Cambridge.
Alongside my industry work, I build independent AI applications for art, real-time 2D and 3D generation, autonomous navigation and financial analysis. I’m also developing a fast browser-based AI library to make advanced models more accessible.
Live 3D, in your browser.
Neural architecture in HD, transforming live.
- Based in
- Zurich, Switzerland
- Languages
- English (C1), German (B1), Russian (native)
Away from the keyboard
- Biking
- Hiking
- Travelling
- Museums
- Chess
- Tennis
- Swimming
- Running
Feel free to reach out to me or message my chatbot by clickingAsk AI
.
Daniil EmtsevAI Research Engineer · Zurich
AI that sees, guides and creates.In real time.
I build machine-learning systems that work in the moment: computer vision and language models that coach surgeons in training at VirtaMed, and generative AI that runs entirely in your browser.
Selected work
Built to work in the moment.
Surgical coaching, reinforcement learning and generative models, each working in real time.
Surgical AI · VirtaMed
Laparoscopic Skills Trainer
Computer vision that referees keyhole-surgery drills, running entirely in the browser.
VirtaMed · Surgical AI
AI alongside the trainee.
From a suturing coach to a surgical AI proctor and real-time guidance for laparoscopic exercises. Explore all 7 demos from my work at VirtaMed.
Suturing coach on an iPad · Davos 2025 demo
The camera watches a suturing pad; the checklist advances by itself and the coach explains each step.
Read the projectSurgical Video AI Analyser
Stage tracking, safety alerts and an "ask the AI" box on a simulated laparoscopic cholecystectomy.
Read the projectStratafix guidance · incision closure
A continuous closure followed from first pass to final cut.
Read the projectPeg transfer
Objects are counted as they cross mid-air from one side to the other and back.
Read the projectExtracorporeal knot
A simple suture through a penrose drain, timed and checked step by step.
Read the project
The work behind the demos
All work
The full catalog.
14 projects and 11 videos, from surgical AI to research papers. Narrow them down by area or topic, or search for a method or a tool.
2videosAI Proctor
A real-time AI coach that watches surgery training and talks you through it.
- LLM & VLM
- Computer Vision
- Real-time
Astro Pilot
A spacecraft that learns to fly, live in your browser.
- Reinforcement Learning
- In-browser AI
- 3D
5videosLaparoscopic Skills Trainer
Computer vision that referees keyhole-surgery drills, running entirely in the browser.
- Computer Vision
- Real-time
- In-browser AI
Generative Realism for Simulators
Generative models that make a simulator look like the operating room.
- Generative AI
- Computer Vision
- Medical
Ultrasound Anatomy Detection
Finding anatomy in ultrasound, from 60% to 90% accuracy.
- Computer Vision
- Medical
- Synthetic data
Camera Pose from a Point Cloud
Where was this photo taken? Matching pixels directly to 3D points.
- Computer Vision
- 3D
- Patent
De Novo Drug Design from Protein Pockets
Reading a protein pocket in 3D and writing a molecule that fits.
- Generative AI
- Drug discovery
- 3D
Dynamic Plane Convolutional Occupancy Networks
Letting the network choose the planes that describe a 3D shape best.
- Computer Vision
- 3D
- Publication
Barcodes of Loss Landscapes
Measuring how hard a neural network's loss surface is to descend.
- Deep learning theory
- Publication
Alzheimer's Brain Changes with GANs
Showing what Alzheimer's disease does to one specific brain.
- Generative AI
- Medical
- Computer Vision
Nothing matches that combination.
Try removing a topic or changing the search.
Journey
From olympiad physics && math to applied AI.
A degree at MIPT in Moscow, computer vision research at ETH Zurich, then medical AI in industry.
2022 – now
Experience
Machine Learning Research Engineer
VirtaMed · Zurich
- Real-time computer vision in PyTorch for a browser app that analyses live camera input, with inference running fully client-side.
- LLM-based guidance and automated report generation integrated into surgical simulators.
- Transformer-based detection of anatomy in ultrasound, from 60% to 90% accuracy.
- A texture-enhancement system for visual realism, and a scalable pipeline that turns 3D simulation output into training data.
- End-to-end MLOps on Azure ML and DVC for reproducible training, versioning and deployment.
- Teaching: AI lectures at VirtaMed, supervising students from Swiss universities, and contributing to Innosuisse surgical proficiency projects.
2019 – 2022
Education
MSc Computational Science and Engineering
ETH Zurich · Zurich
- Focus on robotics. Master’s thesis in the Computer Vision Lab, graded 5.75 out of 6.
- Supported by the ETH Zurich Master’s Scholarship, covering full tuition and living expenses.
Summer 2019
Experience
R&D Software Engineer, Internship
Data Analytics Group · Moscow
- Co-developed a scalable topological method, persistence barcodes, for analysing the loss landscapes of neural networks.
Summer 2018
Experience
R&D Software Engineer, Internship
Amgen Scholars Program · Zurich
- Used generative adversarial networks to characterise structural brain changes of Alzheimer’s disease in MRI scans.
- Presented the method and findings at the Cambridge Amgen Scholars Symposium.
2015 – 2019
Education
BSc Computer Science and Electrical Engineering
Moscow Institute of Physics and Technology · Moscow
- Focus on data science. GPA 9.0 out of 10, graduated with distinction.
- Lecturer at the MIPT Education Center, 2016 – 2017: designed and taught Olympiad courses in electrostatics and geometry, and wrote a problem book of original questions with solutions.
Research
Publications and a patent filing.
Work on 3D vision, the theory of deep learning and medical imaging, with groups at ETH Zurich and in Moscow.
Patent · 2023
A method for determining the 6D pose of a camera used to acquire an image of a scene using a point cloud of the scene and features
International patent application WO2023186262A1
Paper · 2021
Dynamic Plane Convolutional Occupancy Networks
Winter Conference on Applications of Computer Vision (WACV)
Paper · 2023
Barcodes as Summary of Loss Function Topology
Doklady Rossiiskoi Akademii Nauk. Matematika, Informatika, Protsessy Upravleniya, vol. 514, no. 2
Poster · 2018
Studying Alzheimer’s Disease Related Brain Deformations Using Generative Adversarial Networks
Poster, Cambridge Amgen Scholars Symposium
Google Scholar profile* equal contribution
Tools I work with
- Programming languages
- Python
- TypeScript
- C++
- C#
- C
- R
- MATLAB
- Java
- Machine learning
- PyTorch
- TensorFlow
- TensorFlow.js
- OpenCV
- scikit-learn
- NumPy
- Pandas
- 3D and numerics
- Eigen
- Libigl
- Boost
- Unity
- Platform
- Azure ML
- Azure DevOps
- DVC
- Docker
- Google Cloud
- Git
- Bash
Awards and honours
ETH Zurich Master’s ScholarshipFull tuition and living expenses
2019Singapore International Pre-Graduate AwardBioinformatics
2019Abramov Scholarship for Academic ExcellenceMIPT
2016 – 2019All-Russian Math and Physics OlympiadWinner
2015Tournament of Towns Math CompetitionWinner
2013International Math and Physics Olympiad, BelgradeSilver medal
2013
Contact
Let’s talk.
I enjoy working directly with the people who will use the system. If you have a problem where AI has to work in the real world, I would like to hear about it.