Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I am an AI engineer with 16 years of experience designing, scaling, and deploying machine learning and generative AI systems across defense, enterprise, and global consumer platforms. Currently a Senior AI Engineer at Meta, driving large-scale initiatives in LLaMA, RAG, multimodal embeddings, and PyTorch-based distributed training. I helped build internal AI platforms enabling billions of daily predictions and delivered GenAI tools that let product teams experiment with LLMs in production workflows. Previously at Sutherland, I built predictive analytics and NLP solutions for enterprise clients in finance and telecom, deploying fraud detection, churn prediction, and intent classification systems into production environments via APIs and Java services. I began my career at Leidos as a Data Scientist, applying statistical modeling, anomaly detection, and early ML techniques while developing ETL pipelines and C#/.NET dashboards that made model outputs accessible to analysts. I’m known for bridging AI research and engineering execution, with deep expertise in PyTorch, TensorFlow, FAISS, Kubernetes, and modern MLOps practices. I’m adept at mentoring teams, scaling infrastructure, and translating cutting-edge AI into measurable business impact.…I am an AI engineer with 16 years of experience designing, scaling, and deploying machine learning and generative AI systems across defense, enterprise, and global consumer platforms. Currently a Senior AI Engineer at Meta, driving large-scale initiatives in LLaMA, RAG, multimodal embeddings, and PyTorch-based distributed training. I helped build internal AI platforms enabling billions of daily predictions and delivered GenAI tools that let product teams experiment with LLMs in production workflows. Previously at Sutherland, I built predictive analytics and NLP solutions for enterprise clients in finance and telecom, deploying fraud detection, churn prediction, and intent classification systems into production environments via APIs and Java services. I began my career at Leidos as a Data Scientist, applying statistical modeling, anomaly detection, and early ML techniques while developing ETL pipelines and C#/.NET dashboards that made model outputs accessible to analysts. I’m known for bridging AI research and engineering execution, with deep expertise in PyTorch, TensorFlow, FAISS, Kubernetes, and modern MLOps practices. I’m adept at mentoring teams, scaling infrastructure, and translating cutting-edge AI into measurable business impact.

David Natsvlishvili

AI Engineer, Data Scientist, Full Stack Developer, +2





I am an AI engineer with 16 years of experience designing, scaling, and deploying machine learning and generative AI systems across defense, enterprise, and global consumer platforms. Currently a Senior AI Engineer at Meta, driving large-scale initiatives in LLaMA, RAG, multimodal embeddings, and PyTorch-based distributed training. I helped build internal AI platforms enabling billions of daily predictions and delivered GenAI tools that let product teams experiment with LLMs in production workflows. Previously at Sutherland, I built predictive analytics and NLP solutions for enterprise clients in finance and telecom, deploying fraud detection, churn prediction, and intent classification systems into production environments via APIs and Java services. I began my career at Leidos as a Data Scientist, applying statistical modeling, anomaly detection, and early ML techniques while developing ETL pipelines and C#/.NET dashboards that made model outputs accessible to analysts. I’m known for bridging AI research and engineering execution, with deep expertise in PyTorch, TensorFlow, FAISS, Kubernetes, and modern MLOps practices. I’m adept at mentoring teams, scaling infrastructure, and translating cutting-edge AI into measurable business impact.…I am an AI engineer with 16 years of experience designing, scaling, and deploying machine learning and generative AI systems across defense, enterprise, and global consumer platforms. Currently a Senior AI Engineer at Meta, driving large-scale initiatives in LLaMA, RAG, multimodal embeddings, and PyTorch-based distributed training. I helped build internal AI platforms enabling billions of daily predictions and delivered GenAI tools that let product teams experiment with LLMs in production workflows. Previously at Sutherland, I built predictive analytics and NLP solutions for enterprise clients in finance and telecom, deploying fraud detection, churn prediction, and intent classification systems into production environments via APIs and Java services. I began my career at Leidos as a Data Scientist, applying statistical modeling, anomaly detection, and early ML techniques while developing ETL pipelines and C#/.NET dashboards that made model outputs accessible to analysts. I’m known for bridging AI research and engineering execution, with deep expertise in PyTorch, TensorFlow, FAISS, Kubernetes, and modern MLOps practices. I’m adept at mentoring teams, scaling infrastructure, and translating cutting-edge AI into measurable business impact.

Available to hire

I am an AI engineer with 16 years of experience designing, scaling, and deploying machine learning and generative AI systems across defense, enterprise, and global consumer platforms. Currently a Senior AI Engineer at Meta, driving large-scale initiatives in LLaMA, RAG, multimodal embeddings, and PyTorch-based distributed training. I helped build internal AI platforms enabling billions of daily predictions and delivered GenAI tools that let product teams experiment with LLMs in production workflows.

Previously at Sutherland, I built predictive analytics and NLP solutions for enterprise clients in finance and telecom, deploying fraud detection, churn prediction, and intent classification systems into production environments via APIs and Java services. I began my career at Leidos as a Data Scientist, applying statistical modeling, anomaly detection, and early ML techniques while developing ETL pipelines and C#/.NET dashboards that made model outputs accessible to analysts. I’m known for bridging AI research and engineering execution, with deep expertise in PyTorch, TensorFlow, FAISS, Kubernetes, and modern MLOps practices. I’m adept at mentoring teams, scaling infrastructure, and translating cutting-edge AI into measurable business impact.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Work Experience

Senior AI Engineer at Meta

April 1, 2019 - Present

Drove applied AI initiatives within FAIR and product engineering, focusing on PyTorch-based large-scale model training and deployment to support billions of daily predictions. Built a retrieval-augmented generation (RAG) platform on top of FAISS + PyTorch Lightning, powering internal semantic search across Meta's massive knowledge bases. Prototyped and shipped a GenAI experimentation tool using LLaMA models with a React + GraphQL frontend and Python microservices backend — giving product teams a way to plug LLMs directly into user workflows. Improved distributed training and inference workflows on Kubernetes and Ray, making large-scale PyTorch jobs more reliable and faster to ship. Partnered with infra and product teams to scale multimodal embeddings (vision + text) for recommendation and ranking systems. Established MLOps best practices across teams — drift detection, MLflow-based experiment tracking, and custom Grafana dashboards for production LLMs. Mentored engineers and shared

Machine Learning Engineer at Sutherland

February 1, 2012 - April 1, 2019

Designed and implemented end-to-end ML solutions for global enterprise clients in finance, telecom, and customer support, with a focus on predictive analytics, NLP, and anomaly detection. Developed and deployed text classification, sentiment analysis, and intent detection models using scikit-learn, spaCy, and TensorFlow, significantly reducing manual case handling. Built fraud detection and churn prediction systems leveraging time-series forecasting, GLMs, and gradient boosting — improved detection accuracy and reduced false positives. Partnered with engineering teams to productionize ML models through REST APIs, Java-based services, and SQL pipelines, ensuring seamless integration with client applications. Established feature engineering and evaluation frameworks to improve model robustness and reliability in production.

Data Scientist at Leidos

June 1, 2008 - February 1, 2012

Developed statistical and machine learning models for defense and government programs, focusing on predictive analytics, anomaly detection, and risk assessment. Designed and maintained ETL pipelines in Python and SQL to consolidate and process large, heterogeneous datasets for downstream analysis. Applied classical ML techniques (logistic regression, clustering, ensemble methods) to improve intelligence analysis and operational decision-making. Built interactive visualization and reporting tools using C#/.NET, enabling analysts to explore model outputs and trends in real time. Partnered with engineering teams to integrate models into production workflows, balancing accuracy, interpretability, and computational efficiency. Advocated for the early adoption of data-driven approaches, helping teams transition from manual reporting toward automated, AI-enabled insights.