Resume

• Reporting & Widgets team — widget governance, scaling reporting infrastructure, and performance optimisation.
• Driving AI integration across the team’s product surface as part of Monday.com’s AI-first strategy.

• Owned company-wide AI strategy and P&L, prioritising 6-month revenue & cost-saving bets with C-suite.
• Architected distributed embedding & inference clusters (ONNX + Triton) on AWS EKS; auto-scaling and vector-DB sharding cut compute costs by 45%.
• Founded and directed the Data and AI department, selecting and nurturing a team of 12 engineers, boosting company revenue by 20%.
• Created MLOps platform based on AWS SageMaker and RayTune with Terraform, covering all pipeline stages from experimentation to monitoring.
• Orchestrated the development of a multimodal LLM agent for project management, enhancing lead generation by 60%.
• Shipped text2sql and RAG solution from POC to production service.
• Improved CV model inference performance with GStreamer by 40%.
• Centralised monitoring and alerting with Victoria Metrics and Grafana, improving incident response time by 75%.
• Coordinated delivery of 8+ end-to-end data/ML products, working with engineering, product, and design teams.
• Streamlined product development with RICE framework, eliminating 50% of non-essential features and increasing team productivity by 40%.

• Led a cross-functional team of 10 ML and software engineers on CLTV prediction, churn prediction, load forecasting, and recommendation systems.
• Designed and developed a Machine Learning Recommendation Platform serving tens of models to more than 2 million MAUs.
• Built MLOps infrastructure with K8s, Kubeflow, Airflow, and DVC, enabling 6 recommendation model types after hundreds of A/B tests.
• Architected an Airflow DAG Factory enabling data scientists to add new DAGs as YAML configuration with little to no code, reducing data preparation delivery time by 75%.
• Introduced engineering interview flow that raised the bar of candidates while decreasing the interview loop.

AJNA – an AI-powered augmented live video streaming tool that recognises the game flow of sporting events, provides advanced player tracking in live mode, and collects technical data to create new markets.

• Guided a team of 3 in implementing a broadcast video streaming service from a bespoke camera based on Nvidia Jetson; progressed from POC to first production release for live soccer broadcasts.
• Implemented smooth PTZ frame transition algorithm for broadcast.
• Engineered real-time panorama stitching of dual 4K frames using OpenCV on CPU and GPU with CUDA, achieving 40% reduction in processing time.

• Moved the main post-processing video CV pipeline to Google Cloud services.
• Designed and developed the first customer-facing statistics service.
• Improved CV video processing pipeline performance by 54% (99th percentile).

• Migrated data infrastructure from Data Warehouse to Data Lake.
• Implemented real-time data ingest flow with AWS Kinesis, processing 10,000+ data points per second and reducing latency by 60%.
• Developed an NLP articles category classifier, reducing content categorisation time by 70%.
• Improved data infrastructure, reducing service latency 3× from 9s to under 2s.

• Refactored test automation framework, saving 25% maintenance time.
• Trained and mentored 2 new team members.

• QA Engineer, Waze/Alphabet (Jan 2017 – Jul 2017) – led quality assurance on the Waze Carpool Rider app. Cut regression cycle time 2× by switching to checklist test documentation.
• QA Engineer, ex-Cogniance/Star (Sep 2015 – Jan 2017) – cut regression cycle time 2× via risk-based testing strategy. Established QA processes leading to 40% fewer bugs and 30% faster time-to-market.

Master’s thesis: CNNs for network users’ biometric identification.