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

I’m a Machine Learning Engineer with a focus on computer vision and applied LLM integration, delivering end-to-end ML solutions from data pipelines to model training and evaluation. I enjoy turning messy real-world data into robust, scalable systems and collaborating with cross-functional teams to ship impactful AI products. My work blends hands-on research with production-grade engineering: building real-time detection pipelines, evaluating models at scale, and automating reproducible workflows to accelerate deployment and domain adaptation.…I’m a Machine Learning Engineer with a focus on computer vision and applied LLM integration, delivering end-to-end ML solutions from data pipelines to model training and evaluation. I enjoy turning messy real-world data into robust, scalable systems and collaborating with cross-functional teams to ship impactful AI products. My work blends hands-on research with production-grade engineering: building real-time detection pipelines, evaluating models at scale, and automating reproducible workflows to accelerate deployment and domain adaptation.

Iaroslav Aksenkin

Data Scientist, AI Engineer, AI Developer, +2





I’m a Machine Learning Engineer with a focus on computer vision and applied LLM integration, delivering end-to-end ML solutions from data pipelines to model training and evaluation. I enjoy turning messy real-world data into robust, scalable systems and collaborating with cross-functional teams to ship impactful AI products. My work blends hands-on research with production-grade engineering: building real-time detection pipelines, evaluating models at scale, and automating reproducible workflows to accelerate deployment and domain adaptation.…I’m a Machine Learning Engineer with a focus on computer vision and applied LLM integration, delivering end-to-end ML solutions from data pipelines to model training and evaluation. I enjoy turning messy real-world data into robust, scalable systems and collaborating with cross-functional teams to ship impactful AI products. My work blends hands-on research with production-grade engineering: building real-time detection pipelines, evaluating models at scale, and automating reproducible workflows to accelerate deployment and domain adaptation.

Available to hire

I’m a Machine Learning Engineer with a focus on computer vision and applied LLM integration, delivering end-to-end ML solutions from data pipelines to model training and evaluation. I enjoy turning messy real-world data into robust, scalable systems and collaborating with cross-functional teams to ship impactful AI products.

My work blends hands-on research with production-grade engineering: building real-time detection pipelines, evaluating models at scale, and automating reproducible workflows to accelerate deployment and domain adaptation.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

English

Fluent

Work Experience

Machine Learning Engineer at ActionEngine

February 1, 2025 - Present

Built the core training-data pipeline for real-time detection systems (5-15+ classes per deployment), pulling post-review field data from a shared database into reusable datasets for training and evaluation, reducing per-iteration dataset setup from about half a day of manual work to roughly an hour. Contributed to training and evaluation of a production detection model (15 classes), achieving 0.65 mAP on a held-out validation set. Built an internal evaluation tool to surface per-domain failure cases before they reach the detector. Added MLflow experiment tracking and a class-mapping module enabling flexible training configurations across class IDs. Built an ingestion pipeline for Mapillary to bootstrap detection in new regions, closing the domain gap at launch. Automated model container image builds and publishing to the container registry in CI, replacing manual packaging with reproducible builds.

Machine Learning Engineer at ITMO University

October 1, 2021 - December 1, 2025

Arctic ice forecasting framework: designed and implemented an end-to-end forecasting system integrating four heterogeneous historical datasets, achieving 0.07 L1 error and 0.98 SSIM at a 2-year horizon. LLM assistant for urban services: built applied LLM integration layer with multi-provider connectors (predating LangChain), prompt engineering and parameter tuning, agent-tool connectors, and evaluation pipeline using GEval; project GEval improved from 0.30 to 0.83. Neural architecture search framework: rebuilt a non-functional prototype into a modular framework with skip connections in generated architectures; ROC-AUC improved from 0.970 to 0.987 on image classification. Additional LLM work: OpenRouter connector for an in-house LLM tool, benchmarking pipelines for internal LLM project, and LoRA fine-tuning of Llama 3.