Toni Vlaic
Verified Expert in Engineering
Software Developer
Experienced in all phases of the development life cycle, Toni enjoys improving business processes and creating value for his clients. His professional experience ranges from creating REST APIs and web development to enabling his clients to make data-driven decisions using machine learning and data science. Practical and technical knowledge as a developer and a business consultant make Toni extremely capable to work with teams of any size.
Portfolio
Experience
Availability
Preferred Environment
CODE, Vim Text Editor, Git, Bash, MacOS, Linux
The most amazing...
...thing I've coded was a model for credit scoring for a regional bank. The model was a 12% improvement and also presented at a Fintech conference in London 2019.
Work Experience
Analytics Engineer
205 Data Lab
- Leveraged a vast client database of company information to create a productized SFDC plugin with KPI descriptors using DBT and Python orchestrated by Airflow.
- Performed deep-dive analysis and data cleanup and consolidation of millions of records between clients' product and SFDC instance for easier BI, reporting, and decision making.
- Implemented Looker explores to expose BigQuery data to non-technical users in an easy to consume way.
Data Scientist/Engineer
Bonsai D.O.O
- Served as a data scientist on the data science team.
- Created an ML-powered ticket automation plugin for Zendesk and Service Shift platform that auto-assigns tickets and offers automatic response capabilities with predefined macros to reduce ticket response time and improve customer service.
- Developed a deep learning model and API for a local car sales company that enables their clients to create professional car photos for their sale ad directly in their parking lot; it improved efficiency and shortened the process of ad creation.
- Worked on the implementation and design of ETL pipelines and data modeling of a DWH for a global coffeehouse company to serve as a single source of truth for reporting and analytics.
Data Scientist
Athena Analytics
- Transitioned existing models to LightGBM counterparts to improve the handling of missing values.
- Worked closely with the engineers and manager to spec out a road map for a career recommendation solution as a future improvement to the system.
- Performed research on existing data sets and implemented a recommendation engine in collaboration with psychologists to suggest the best career fit based on a person's psychological and academic traits.
- Migrated code to dockers and implemented Linux services to make the solutions easier to scale and manage.
Data Analyst
Toptal Client
- Contributed to deep-dives and reporting as a product data analyst.
- Collaborated closely with the product managers and engineers to translate business insights into actions and helped define business metrics, product roadmaps, and quarterly OKRs.
- Created and supported multiple dashboards on deep-dive using SQL, Python, and Mode Analytics.
- Presented actionable insights using Mode Analytics dashboards to managers and engineers to optimize revenue, pricing, and reduce fraudulent activity.
- Implemented models in the ETL pipeline that transforms the available data into a standardized single source of truth.
Senior Analyst
CantabPI
- Led a team during the creation of a marketing optimization tool for a Fortune 50 pharma company. The resulting model was able to quantify the impact of each action on sales and optimize action prioritization and resource management.
- Developed a behavioral credit scoring model for a regional bank to reduce credit risk and take advantage of the new PSD2 directive. The model achieved 12% better results than the previous one and was presented at a fintech conference in London, 2019.
- Created a capacity management model for a big regional hotel chain owner to predict cancellations and reservations for making overbooking decisions. Hotels filled 100% capacity during prime time while the best previously achieved was 96%.
R&D Engineer
Span
- Worked on the computer vision research and development team.
- Developed face recognition and age/emotion detection using deep learning to analyze customers and cross-reference their extracted demographic data with bought consumer goods to gain further insight into behavioural patterns.
- Created a signature verification model to detect if a person's signature is genuine (in case they forget which one was used) to speed up the process of approving requests and reduce manual labor.
- Created a solution that analyzes the global DNS traffic of a big client with over 300 million requests daily. The solution used outlier detection and statistics to prevent DNS tunneling attacks.
Software Developer
Span
- Worked in the cloud solutions team.
- Collaborated on the creation of an internal survey creator and management website. Developed the website back end in C# and MS SQL which was deployed on Azure cloud services.
Experience
Trimmy - Script Creation Application
Core member of a team competing in AppStart contest responsible for OCR and back-end logic including the export functionalities.
Technologies: Python, Django, Android, Heroku, OpenCV, GoogleAPI
Robust ML Challenge
https://github.com/Mungosin/MozgaloThe developed ensemble of classifiers can process a batch of up to 16 images within 1 second on a regular CPU and has 94% F1 macro metric making it extremely good across all classes.
Technologies: Python, Keras, OpenCV
Extracting Deep Features for Image Recommendation
https://github.com/Mungosin/AVSPThe resulting engine was later to develop a demo recommendation engine application that received the Rector's award for Individual scientific work.
Technologies: Python, Tensorflow, Django, Heroku
Kaggle Expert Competitor
https://www.kaggle.com/mungosImage Segmentation problem - Detection of Nuclei in Cell Images
Result: 91/3634 - Top 3%
Technologies: Python, Keras, OpenCV
https://www.kaggle.com/c/traveling-santa-2018-prime-paths
Traveling Salesman Problem
Result: 42/1871 - Top 3%
Technologies: Python, Numba
https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection
Detecting fraudulent click traffic
Result: 308/3946 - Top 8%
Technologies: Python, LightGBM, Scikit-Learn, Pandas
https://www.kaggle.com/c/home-credit-default-risk
Developing a Credit Scoring model
Result: 459/7190 - Top 7%
Technologies: Python, LightGBM, Scikit-Learn, Pandas
Web Scraper
Technologies: Python, Selenium
Education
Master's Degree in Computer Science
University of Zagreb - Zagreb, Croatia
Bachelor's Degree in Computer Science
University of Zagreb - Zagreb, Croatia
Skills
Libraries/APIs
Pandas, NumPy, Matplotlib, Scikit-learn, Keras, REST APIs, SQLAlchemy, PyTorch, OpenCV, TensorFlow
Tools
Jupyter, GitHub, Git, Apache Airflow, Plotly, BigQuery, Looker, Tableau
Languages
Python, Snowflake, SQL, Bash
Frameworks
Django, LightGBM, Django REST Framework, Flask
Paradigms
Data Science, ETL, Object-oriented Programming (OOP), Continuous Integration (CI), Continuous Deployment
Storage
PostgreSQL, Microsoft SQL Server, MySQL
Platforms
Jupyter Notebook, Databricks, Azure Synapse, Linux, OS X, Azure, Docker
Other
Machine Learning, Google BigQuery, Data, Data Analytics, APIs, Back-end, Deep Learning, Data Visualization, Data Preprocessing, Data Analysis, Data Build Tool (dbt), Mode Analytics, API Integration, Azure Data Factory, Prefect, Web Development
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring