The Wire
← back to the wire

Yury Lapin

Director Data Science at Presight.ai a G42 Company
17.3 yrs experienceComputer Software
Mar 2024

The move

MovePresight.ai → Presight
TitlePrincipal Scientist → Director Data Science
LocationAbu Dhabi → Abu Dhabi
Career arcRooted in UAE★
Cross-border moveNo — same country
Native languageEnglish, Russian
Years in UAE14.0
Total experience17.3 years
Presight headcount501-1,000 employees
Presight industryIT Services and IT Consulting
Presight HQAbu Dhabi, UAE
Presight typePublic Company

Career arc

  1. Director Data ScienceMar 2024 — Present
    Presight· Abu Dhabi Emirate, United Arab Emirates· Full-time
  2. Principal ScientistJan 2023 — Mar 2024
    Presight.ai· Abu Dhabi, Abu Dhabi Emirate, United Arab Emirates· Full-time
  3. Lead Data Scientist / Solution ArchitectNov 2016 — Mar 2023
    G42· Abu Dhabi, United Arab Emirates· Full-time
  4. Solution ArchitectNov 2015 — Oct 2016
    UAE Govt. Project· Abu Dhabi
  5. Team Leader/Senior Technical ConsultantAug 2012 — Oct 2015
    ATS Group· Abu Dhabi
  6. Server Side DeveloperApr 2012 — Jul 2012
    ATS Group· Abu Dhabi
  7. Senior .NET Developer/Team LeaderApr 2011 — Apr 2012
    TopS Business Integrator· Moscow, Russian Federation
  8. .NET DeveloperOct 2008 — Apr 2011
    DataArt· Voronezh, Russina Federation

Education

Languages

  • French · Elementary
  • English · Native or bilingual
  • Russian · Native or bilingual

Top skills

TensorFlowObject-Oriented Programming (OOP)LDAPyTorchPCARust

About

• 14 years of affluent hands-on experience as Data Scientist/Senior Developer/Team Leader/Architect • Deep learning models development and fine-tuning using Pytorch, Keras and Tensorflow for various tasks on imagery/video and audio data with Python. • Entropy guided active learning applications on top of reach semantic text-to-image or video embeddings (like CLIP, X-CLIP). • Production deployment with Airflow and TensorRT inference Server (Triton). • Stochastic Weights Averaging Gaussian (SWAG) approach to lower epistemic bias on the image recognition model trained with synthesised data. • Video anomaly detection based on weakly-supervised learning on extracted embeddings from video segments using R(2+1)D based model architecture. • Discriminative models for langu

Share this move

View profile on LinkedIn