Open to opportunities

Valentin Lapparov

@valentinlapparov

Message

I build scalable machine learning systems for search, computer vision, and recommendation products.

France

Message

At Linkup, I fine-tune LLMs for an AI-powered web search platform for agents and build Vespa search ranking systems that improved relevance while cutting search and inference costs by 70%.

Previously, I built scalable image-processing systems at OCUS, reducing infrastructure costs by 50% and halving inference time. I also co-founded CreoAI, led machine learning for personalized AI art and recommendations, and developed streaming anomaly-detection pipelines at PacketAI that reached a 96% F-score and ran five times faster.

Experience

Work history, roles, and key accomplishments

Current

Founding Machine Learning Engineer

Current

Linkup

Feb 2025 - Present (1 year 6 months)

Contributed to an AI-powered web search platform for AI agents by fine-tuning LLMs for search-related tasks within an agentic framework, improving retrieval quality, factual accuracy, and cost-performance tradeoffs in production. Built and optimized a Vespa-based search index and ranking stack, implementing state-of-the-art ranking functions and embedding techniques that improved relevance while r

LLM Fine tuning Vespa Ranking Embeddings Spark Databricks Data Pipelines

Senior Machine Learning Engineer

OCUS

Sep 2023 - Feb 2025 (1 year 5 months)

Architected and built from scratch a highly scalable Machine Learning system for image processing. Reduced company infrastructure costs by 50% while halving inference time, and designed and implemented image processing pipelines that leverage cutting-edge Computer Vision models.

Machine Learning Computer Vision Image Processing Scalable Systems

Co-Founder and Machine Learning Team Lead

CreoAI

Aug 2022 - Dec 2023 (1 year 4 months)

Led the development of a social network that allowed users to create and explore personalized art generated by AI. Architected the Machine Learning system for image generation tailored to each user's personality and preferences, and developed a recommendation system that curated individualized content feeds for users, optimizing engagement and user retention.

Machine Learning Image Generation Recommendation Systems Team Leadership

Founding Machine Learning Engineer

PacketAI

Jan 2020 - Aug 2023 (3 years 7 months)

Developed from scratch a highly scalable and distributed end-to-end Machine Learning pipelines for anomaly detection in streaming data, reaching 96% of f-score while making the pipeline 5 times faster. Managed a team of ML engineers and owned the end-to-end lifecycle of a product, from development to deployment and maintenance.

Machine Learning Anomaly Detection Streaming Data Kafka Elasticsearch MongoDB AWS Kubernetes

Data Science Intern

Ubisoft

Apr 2019 - Jul 2019 (3 months)

Developed a set of methods that enhanced company's ability to measure the impact of CRM campaigns. Built predictive models that increased customer retention and led to discovery of users insights.

Data Science predictive modeling CRM Customer Retention

Data Science Intern

Microsoft

May 2018 - Aug 2018 (3 months)

Built behavioural clusters of To-Do App users and found relevant insights which have led to metrics and app improvements. Developed correlation between customer's behavioural clusters and retention, and established the distribution in task completion time of To-Do data.

Data Science Retention Analysis

Education

Degrees, certifications, and relevant coursework

HEC Paris

Master of Science, Data Science

2017 - 2019

M. Sc. Data Science for Business: Joint Degree with École polytechnique. Subjects studied: Regression, Deep Learning, Database Management, Time Series for Finance.

École Polytechnique

Master of Science, Data Science

2015 - 2018

Ingénieur Polytechnicien Program (M. Sc. in Data Science). №1 Best French Engineering school. Subjects studied: Statistics, Mathematics, Machine Learning, Optimisation.