Dawid is available for hire
Hire DawidDawid Smoleń
Verified Expert in Engineering
Machine Learning Engineer and Software Developer
Kraków, Poland
Toptal member since September 15, 2021
Bio
Dawid has successfully delivered 30+ machine learning projects, building scalable systems grounded in MLOps best practices. With deep expertise in cloud-native technologies, he automates the entire ML lifecycle, from data gathering to CI/CD pipelines and continuous training. Recently, Dawid has been applying this expertise to LLM-based solutions at scale, helping clients unlock cutting-edge AI capabilities.
Portfolio
Freelance
Python, Scikit-learn, Docker, Amazon Web Services (AWS)...
Sinch
Kubernetes, Google Kubernetes Engine (GKE), Docker, Helm, GitOps...
Toptal
Python, PyTorch, Machine Learning Algorithms, Machine Learning...
Experience
- Scikit-learn - 10 years
- Machine Learning Automation - 6 years
- PyTorch - 6 years
- Generative Pre-trained Transformers (GPT) - 6 years
- CI/CD Pipelines - 5 years
- Azure - 4 years
- Kubeflow - 2 years
- Agentic AI - 1 year
Work Experience
ML Consultant | MLOps Engineer
2018 - PRESENT
Freelance
- Deployed modeling services to Kubernetes clusters, Amazon EKS, and Google Kubernetes Engine (GKE).
- Introduced tracking servers to the existing projects to improve the observability of a model and understanding of a problem.
- Developed an end-to-end solution from data investigation to a deployed model that monitors daily statistics and business metrics regarding user experience in eCommerce.
- Consulted an ECG-related company from Latin America. Helped with the design and implementation of crucial Holter analysis steps.
- Prepared NFT market analysis tools based on machine learning traits valuation.
- Prepared a deduplication service for a real estate website scraper.
- Acted as a data science trainer for two training companies and conducted training for around seven teams from various enterprises.
- Transformed, together with a team, the way Médecins Sans Frontières understands their communication works in the wild.
Technologies: Python, Scikit-learn, Docker, Amazon Web Services (AWS), Digital Signal Processing, ECG, Training, Data Science, Jupyter Notebook, Artificial Intelligence (AI), Machine Learning, Regression Modeling, Classification Algorithms, Kubeflow, Kubernetes, CI/CD Pipelines, Machine Learning Operations (MLOps), Data Scraping, Data Engineering, Front-end, Data Analysis, Non-fungible Tokens (NFT), Retrieval-augmented Generation (RAG), Large Language Models (LLMs), Large Language Model Operations (LLMOps), SQL, MongoDB, APIs, AI Agents, Continuous Integration (CI), Machine Learning Automation, Azure, REST APIs, DevOps, Temporal Cloud, Events, FastAPI, Amazon EC2, Amazon S3 (AWS S3), ML Pipelines, Infrastructure as Code (IaC), Model Deployment, Monitoring, Observability Tools, DSP, Orchestration, Temporal
MLOps Engineer
2022 - 2025
Sinch
- Introduced the best MLOPS practices at Chatlayer, managing thousands of models in production. Maintained them and also significantly optimized the costs and speed.
- Drove the adoption of AI solutions across multiple departments at Sinch, including automated campaign analytics, anomaly detection in messaging systems (email and SMS), system integration, and the development of unified standards.
- Participated in a few LLM and agentic projects running at a large scale, transforming industries.
- Utilized cutting-edge technologies, including GitOps and event-based architecture, as well as workflow automation using Temporal and Argo Workflows.
- Migrated massive projects between popular cloud providers.
- Created an anomaly detection and prediction system for the mailing industry. It processes 2-3 billion emails daily.
- Enhanced observability by adding tools at multiple levels.
- Implemented a custom high-density model deployment platform (a Kubernetes engine for 2,000 models).
- Implemented a highly distributed and responsive compliance review system on Temporal.io. Everything is parallel and tolerant; an optimized DAG processes the input data as it arrives and assesses the accordance with hundreds of rules, LLM-based.
Technologies: Kubernetes, Google Kubernetes Engine (GKE), Docker, Helm, GitOps, Large-scale Projects, Artificial Intelligence (AI), Data Engineering, Python, LangChain, Prompt Engineering, OpenAI, Amazon Web Services (AWS), SQL, MongoDB, Vector Databases, LangGraph, Generative Artificial Intelligence (GenAI), APIs, AI Agents, Continuous Integration (CI), Machine Learning Automation, Google Cloud, REST APIs, DevOps, Temporal Cloud, Argo Workflows, Events, FastAPI, Databricks, ML Pipelines, Terraform, Model Deployment, Monitoring, Observability Tools, Agentic AI, Orchestration, Temporal
Machine Learning Consultant
2021 - 2022
Toptal
- Introduced MLOps design patterns, including pipelines, observability tools, and monitoring solutions, and deployment to a Kubernetes cluster.
- Built a PoC for an automatic real estate valuation system. Set up the codebase and pipelines, conducted research, and developed a few competitive prototypes.
- Participated in establishing a team by hiring and training members who later took over the project.
Technologies: Python, PyTorch, Machine Learning Algorithms, Machine Learning, Amazon SageMaker, REST APIs, DevOps, FastAPI, ML Pipelines, Model Deployment, Monitoring, Observability Tools
Machine Learning Engineer
2020 - 2021
Grape Up
- Developed an end-to-end deep learning automotive project together with full automation (CI, CD, and CT) and infrastructure. Worked on machine learning best practices using modern tools and solutions.
- Created POCs and demos in machine learning and data science areas, together with simple UI demos and an API first approach.
- Built a VIN recognition system. https://medium.com/grapeup/leveraging-ai-to-improve-vin-recognition-how-to-accelerate-and-automate-operations-in-the-12eac5286b1d.
- Created a blog post. Building Intelligent Document Processing Systems: https://grapeup.com/blog/introduction-to-building-intelligent-document-processing-systems/.
Technologies: Azure, Convolutional Neural Networks (CNNs), Deep Learning, Dataiku, Python, Digital Signal Processing, DevOps, Machine Learning Operations (MLOps), REST APIs, React, Deep Neural Networks (DNNs), Metaflow, Data Science, Jupyter Notebook, Artificial Intelligence (AI), Predictive Modeling, Machine Learning, Amazon Web Services (AWS), Data Scraping, APIs, Continuous Integration (CI), Machine Learning Automation, ETL, PDF Scraping, Data Extraction, FastAPI, Amazon EC2, Amazon S3 (AWS S3), ML Pipelines, Model Deployment, Monitoring, Observability Tools
Deep Learning Engineer
2017 - 2018
Lekta
- Created a library for users' intent classification that employs industry best practices to make predictions millions of times a month in a real-time, demanding environment.
- Developed a novel speech recognition system based on state-of-the-art papers that beat the current market in some areas in terms of accuracy or performance.
- Researched numerous topics in the areas of speech recognition, voice-based gender recognition, intent classification, sentence representation, and text representation.
- Developed machine learning algorithms for both voice bots and chatbots.
Technologies: C++, Audio Processing, Digital Signal Processing, Python, PyTorch, TensorFlow, Deep Learning, Speech Recognition, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Scikit-learn, Data Science, Jupyter Notebook, Chatbots, Artificial Intelligence (AI), Predictive Modeling, Amazon Web Services (AWS), Machine Learning, Data Scraping, Data Engineering, Machine Learning Automation, ML Pipelines, Model Deployment, Monitoring, Observability Tools, DSP
Machine Learning Engineer
2016 - 2017
Aspel SA
- Created a brand new QRS detector tested on many benchmarks and real-world monitoring tests.
- Developed clustering algorithms that can efficiently cluster long Holter monitor tests, focusing on user experience.
- Developed embedded resampling algorithms for ECG devices.
- Contributed to QRS morphology classifiers that highly improved the work of doctors and met AMA standards.
- Helped develop user experience-related algorithms that simplify the work of the doctors and technicians.
Technologies: C++, Python, MATLAB, Scikit-learn, SciPy, Artificial Intelligence (AI), Predictive Modeling, Classification Algorithms, Regression Modeling, Data Analysis, Model Deployment, Monitoring, Observability Tools, Anomaly Detection, DSP
NLP Engineer
2015 - 2016
WitKom – Virtual Translator of Sign Communication
- Developed the first Polish to Polish Sign Language translation system on the language level.
- Built the first Polish Sign Language to Polish translation system on the language level using Seq2Seq models.
- Created huge artificial datasets for sign languages based on heuristics, rules, and DL technology.
Technologies: Python, Natural Language Toolkit (NLTK), Deep Learning, Sequence Models, TensorFlow, Data Analysis, Monitoring
Project History
Gomrade — Play Go Against AI on a Real, Physical Board
https://github.com/smolendawid/GomradeThis repository allows you to play Go with strong AI on a real board. Gomrade analyses the board state from an image using a computer camera and answers the AI moves using a synthesized voice. The example video of Gomrade in action is under development.
Speech Representation and Exploration Notebook
https://www.kaggle.com/davids1992/speech-representation-and-data-explorationThis is one of the top 15 Kaggle notebooks ever, with more than 100,000 views. I introduced a few basic concepts about speech representation and performed data analysis looking for the most interesting examples from the dataset.
The Simplest Python Cache for Data Scientists
https://github.com/smolendawid/cachaThe simplest Python cache for data scientists.
Contrary to many other tools, cacha boasts the following features:
• It is used at the function call, not the definition. Many packages implement the @cache decorator that has to be used before the definition of a function that is not easy enough to use.
• It stores the cache on disk, which means you can use the cache between runs. This is convenient in data science work.
Contrary to many other tools, cacha boasts the following features:
• It is used at the function call, not the definition. Many packages implement the @cache decorator that has to be used before the definition of a function that is not easy enough to use.
• It stores the cache on disk, which means you can use the cache between runs. This is convenient in data science work.
Drifting – The Most Flexible Drift Detection Server
https://github.com/sign-ai/driftingThe most flexible Drift Detection framework for everyone! Python-first, API-first, user-friendly, and open-source!
PYTHON-FIRST
Communicate with the Drift Detection server using a super simple Python client. No additional management needed!
EASY INTEGRATIONS
Using drifting is simple thanks to standardized, ML server-based integrations like Kafka, OpenAPI, and gRPC.
FLEXIBLE
One server for managing many models, projects, versions, and features without any further tools.
STATE-OF-THE-ART
An open-source project built upon the top-tier libraries—Alibi Detect, ML server, and more!
PYTHON-FIRST
Communicate with the Drift Detection server using a super simple Python client. No additional management needed!
EASY INTEGRATIONS
Using drifting is simple thanks to standardized, ML server-based integrations like Kafka, OpenAPI, and gRPC.
FLEXIBLE
One server for managing many models, projects, versions, and features without any further tools.
STATE-OF-THE-ART
An open-source project built upon the top-tier libraries—Alibi Detect, ML server, and more!
My blogging
https://signai.substack.com/I am trying to share my thoughts that I find valuable or unique, on both Medium and Substack and Medium:
- https://medium.com/@smolendawid - migration of a few blogposts from my old personal blog
- https://signai.substack.com/ - more up-to-date blogposts.
- https://medium.com/@smolendawid - migration of a few blogposts from my old personal blog
- https://signai.substack.com/ - more up-to-date blogposts.
Education
2019 - 2021
PhD in Electrical and Electronics Engineering
AGH University of Science and Technology - Cracow, Poland
2011 - 2016
Master's Degree in Acoustical Engineering
AGH University of Science and Technology - Cracow, Poland
Certifications
AUGUST 2021 - PRESENT
ML Practitioner
Dataiku
AUGUST 2021 - PRESENT
Core Designer
Dataiku
SEPTEMBER 2017 - PRESENT
Machine Learning
Coursera
SEPTEMBER 2017 - PRESENT
Neural Networks and Deep Learning
Coursera
SEPTEMBER 2017 - PRESENT
Structuring Machine Learning Projects
Coursera
SEPTEMBER 2017 - PRESENT
Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization
Coursera
Skills
Libraries/APIs
Scikit-learn, PyTorch, TensorFlow, REST APIs, Node.js, React, OpenCV, Keras, SciPy, Natural Language Toolkit (NLTK), Pandas
Tools
Observability Tools, Google Kubernetes Engine (GKE), MATLAB, Helm, Amazon SageMaker, Terraform
Languages
Python, C++, SQL
Paradigms
Continuous Integration (CI), DevOps, Anomaly Detection, ETL
Platforms
Azure, Kubeflow, Jupyter Notebook, Amazon Web Services (AWS), Kubernetes, Temporal Cloud, Dataiku, Docker, Cloud Native, Databricks, Amazon EC2
Storage
Google Cloud, Amazon S3 (AWS S3), MongoDB
Frameworks
Metaflow, LangGraph
Other
Machine Learning Automation, Audio Processing, Natural Language Processing (NLP), Deep Neural Networks (DNNs), Deep Learning, Machine Learning, Sequence Models, Machine Learning Operations (MLOps), ECG, Data Science, Artificial Intelligence (AI), CI/CD Pipelines, Generative Pre-trained Transformers (GPT), Large Language Model Operations (LLMOps), APIs, FastAPI, ML Pipelines, Model Deployment, Monitoring, Temporal, Workflows Orchestration, GitOps, Data Scraping, Data Engineering, Data Analysis, Large Language Models (LLMs), Prompt Engineering, OpenAI, Argo Workflows, Infrastructure as Code (IaC), Agentic AI, Acoustics, Digital Signal Processing, Speech Recognition, Convolutional Neural Networks (CNNs), Training, Chatbots, Predictive Modeling, Regression Modeling, Classification Algorithms, Large-scale Projects, Front-end, Non-fungible Tokens (NFT), Acoustical Engineering, Retrieval-augmented Generation (RAG), IT Project Management, Lecturing, LangChain, Vector Databases, Generative Artificial Intelligence (GenAI), AI Agents, Machine Learning Algorithms, Events, PDF Scraping, Data Extraction, DSP, Cloud, Orchestration
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