Marko is available for hire
Hire MarkoMarko Kostiv
Verified Expert in Engineering
Machine Learning and Software Developer
Lviv, Lviv Oblast, Ukraine
Toptal member since November 23, 2021
Bio
Marko is an AI engineering manager and PhD researcher with 10+ years in software and 7+ years in applied AI. He specializes in real-time computer vision, edge and cloud AI, and generative models. Marko has led teams delivering mission-critical and enterprise AI systems, bridging cutting-edge research with production-grade deployment.
Portfolio
Linearity
Computer Vision, Generative Artificial Intelligence (GenAI)...
Kuna
PyTorch, Amazon Web Services (AWS), Python, C++, Object Detection...
SoftServe
Machine Learning, Deep Learning, PyTorch, Google Cloud Platform (GCP)...
Experience
- Artificial Intelligence (AI) - 8 years
- Python - 8 years
- PyTorch - 7 years
- Amazon Web Services (AWS) - 6 years
- Generative Artificial Intelligence (GenAI) - 4 years
- Computer Vision - 4 years
- Core ML - 2 years
- C++ - 2 years
Preferred Environment
Google Cloud Platform (GCP), Amazon Web Services (AWS), Python, Computer Vision, PyTorch, Machine Learning
The most amazing...
...thing I’ve developed is an AI system used to count and detect plants using drone images for over 400,000 hectares of sunflower and cornfields.
Work Experience
AI Lead
2021 - 2025
Linearity
- Built ML expertise in the company from scratch. Organized interview processes, data collection, annotation pipelines, and release processes.
- Launched the first ML features used by hundreds of thousands of users every month, running on mobile devices, iPads, and laptops.
- Managed a team that launched the first generative artificial intelligence (GenAI) features in early 2023. We built the entire cloud infrastructure for inference from scratch.
Technologies: Computer Vision, Generative Artificial Intelligence (GenAI), Convolutional Neural Networks (CNNs), Artificial Intelligence (AI), Data Science, PyTorch, Core ML, Deep Learning, Machine Learning, Large Language Models (LLMs), Optical Character Recognition (OCR), Data Annotation, ONNX Runtime, NVIDIA TensorRT
Senior Machine Learning Engineer
2021 - 2021
Kuna
- Launched data collection and annotation pipelines from scratch.
- Sped up the model inference by six times while reducing the error rate by 15%.
- Implemented pipelines for model training and end-to-end evaluation.
Technologies: PyTorch, Amazon Web Services (AWS), Python, C++, Object Detection, Convolutional Neural Networks (CNNs), Artificial Intelligence (AI), Data Science, Computer Vision, Deep Learning, Machine Learning, Data Annotation, ONNX Runtime, NVIDIA TensorRT
Lead Machine Learning Engineer
2018 - 2021
SoftServe
- Led pre-sale and consulting activities for Fortune 500 companies.
- Led the creation of the active-learning pipeline for semantic segmentation of 3D seismic data from the PoC phase to the first major release, using PyTorch, GCP, Azure Machine Learning, and SEGY. Grew the engineering team from three to twelve members.
- Led the development of instance segmentation of roof facets and scoped and organized annotation flow using PyTorch and GCP.
Technologies: Machine Learning, Deep Learning, PyTorch, Google Cloud Platform (GCP), Convolutional Neural Networks (CNNs), Artificial Intelligence (AI), Data Science, Computer Vision, TensorFlow, Optical Character Recognition (OCR), Geospatial Data, Data Annotation, ONNX Runtime, NVIDIA TensorRT
Research Intern
2018 - 2018
Smith School of Business at Queen's University
- Developed a CLV prediction engine for credit limit optimization while participating in a joint internship at Queen's University and Scotiabank.
- Implemented survival models for customer lifetime prediction.
- Implemented revenue prediction models using deep learning models and a Markov chain.
Technologies: Machine Learning, Artificial Intelligence (AI), Data Science, PyTorch, Deep Learning, TensorFlow
Lead Software Engineer
2017 - 2018
N-iX
- Reverse-engineered a binary communication protocol for MFi and BLE devices.
- Designed and implemented architecture for iOS and Android applications.
- Led an engineering team of four people and conducted code reviews.
Technologies: Swift, iOS, Android, Core ML
Software Engineer
2014 - 2017
Vakoms
- Led the refactoring initiative of a healthcare app with thousands of DAUs.
- Built a voice IP application for emergency healthcare calls.
- Developed and delivered an image editing toolchain in the social media app.
Technologies: iOS, Swift, C++, Core ML
Experience
AI Backgrounds at Linearity
https://www.linearity.io/blog/linearity-stable-diffusion/At Linearity, I spearheaded the research and development of an AI-powered background generation feature based on diffusion models. Unlike off-the-shelf solutions, the system was custom-trained and fine-tuned to align with creative workflows and deliver outputs that met professional design standards.
I designed the full pipeline—from dataset curation and training to inference optimization and cloud deployment. The model was adapted through targeted training and prompt-conditioning strategies, enabling controllable, high-quality background generation at scale. To ensure production readiness, I implemented inference acceleration, quality filtering, and scalable orchestration in the cloud.
Collaboration with product and design teams ensured the system integrated seamlessly into the Linearity platform, giving thousands of users instant access to unique, custom-generated backgrounds.
I designed the full pipeline—from dataset curation and training to inference optimization and cloud deployment. The model was adapted through targeted training and prompt-conditioning strategies, enabling controllable, high-quality background generation at scale. To ensure production readiness, I implemented inference acceleration, quality filtering, and scalable orchestration in the cloud.
Collaboration with product and design teams ensured the system integrated seamlessly into the Linearity platform, giving thousands of users instant access to unique, custom-generated backgrounds.
Heyoh Camera for Mac
This project was awarded a 3rd place at the PyTorch Annual Hackathon. Heyoh is a virtual camera for Zoom and Google Meet for macOS. Heyoh augments live video in real-time by recognizing hand gestures and smiles and shows effects to highlight them to other video participants. Heyoh works with any video device, either physical or virtual, and since we do everything on your computer, your video calls and data always stay safe.
Education
2024 - 2025
PhD in Intelligent Systems
Ukrainian Catholic University - Lviv, Ukraine
2017 - 2019
Master's Degree in Data Science
Ukrainian Catholic University - Lviv, Ukraine
Certifications
JUNE 2017 - PRESENT
IELTS
British Council
Skills
Libraries/APIs
PyTorch, TensorFlow
Languages
Python, Swift, C++
Frameworks
Core ML
Platforms
Amazon Web Services (AWS), Google Cloud Platform (GCP), iOS, Android
Other
Data Science, Computer Vision, Machine Learning, Deep Learning, Object Detection, Convolutional Neural Networks (CNNs), ONNX Runtime, NVIDIA TensorRT, Optical Character Recognition (OCR), Data Annotation, Drone Photography & Videography, Large Language Models (LLMs), Geospatial Data, English, Generative Artificial Intelligence (GenAI), Artificial Intelligence (AI), Cloud
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring