АО «Bereke Bank»

Risk models
- Significantly enhanced credit risk accuracy and process reliability by developing high-performing CatBoost and LGBM risk models, tailored for diverse financial products, leading to an increased approval rate without raising the false-negative rate.
- Enhanced model performance by incorporating reject inference data, applying advanced feature selection, and optimizing the overall process.

Enhancing model monitoring pipeline
- Improved the sensitivity of the classification model monitoring pipeline by implementing methods of detecting silent model failures through Data Drift measurement and Confidence-Based Performance Estimation.
- Enhanced the productivity of the regression model monitoring pipeline by discovering and implementing Model Performance Predictor.
- Boosted the performance of newly developed models by introducing and integrating probability calibration techniques into the team’s development pipeline.
- Helped the team to accelerate model development and implementation by contribution to development of an advanced company-based toolkit development.

Client's behavior prediction model
- Enabled the successful re-launch of a product with a targeted client proposal sample that drove strong initial engagement by developing a look-alike model using a KNN classifier to predict client behavior from historical transaction data.

Public speaking
- Presented at Citix Meetup 2023 on the theoretical foundations of Model Performance Prediction and Data Drift Analysis
- Delivered a talk at BeeTech Conference 2024 on practical implementation of advanced model monitoring methods.

Work with team
- Collaborated with the team to hire top-tier professionals by conducting technical interviews focused on math, statistics, Python, SQL and core machine learning.
- Helped newbies to grow by mentoring them and leading technical meetups. Provided mentorship to interns, supporting their development and integration into the team.

Senior Data Scientist

Sberbank

Product: Credit cards, issued and delivered to front offices.

Shipment calendarization:
- Discovered a strategy of cost reduction on shipment sending process by conducting exploratory historical data analysis and describing logistic patterns on Python.
- Build a linear-programming model of shipments calendarization with potential 7.5% cost-reducing result achieved by selecting optimal shipment sending days based on discovered logistic patterns. 

Product demand model:
- Ensured sustained material provision and increased precision of planning for 20% by conducting a hundred-product-wide research on credit card demand historical data and developing a model on Python based on time series techniques - SARIMAX and multi-layer filtration.
- Reduced idle rate on production due to raw material shortages for 60% and hand labor amount for 80%  by building a model of raw material supplies between 8 production centers on Python based on linear programming algorithms. This allowed a team to handle a crisis situation with multiple shortages.

Production model:
- Enhanced prediction accuracy for 10% and made the load planning process more insightful by switching previously developed Excel-based model completely to Python and using more detailed data with linear programming algorithms.
- Calculated precise production machines speeds profiles by conducting EDA with several gigabytes of raw production log data. These profiles were used in the new production model as a core set of parameters.

Increasing report effectiveness:
- Significantly increased stability of logistic service status dashboard preparation process and established warning tools by switching Excel-based tools to Postgre SQL using dBeaver.
- Provided more effective and less vulnerable delivery data collection and preparation process by leading a development project on AB-Initio with a team of 3 developers.
- Led a quarter-long education process for a team of 3 developers on building AB-Initio graphs.

Middle Data Scientist

Product: Credit cards, issued and delivered to front offices.

Cost reduction:
- Reduced the average amount of extra production for 5% by conducting an EDA about unclaimed credit card amounts combined with client’s behavioral data and building a production demand model based on linear regression on Python.
- Reduced production costs for 9,6% a year and simplified decision-making process by creating an Excel- and VBA-based production model which builds machines relocation scenarios based on issuing demand forecasts.

Report automatization:
- Enhanced accuracy for 15% and reduced for 10 hours a month hand labor amount of delivery service monitoring process by developing metrics calculation and visualization tools based on Power-BI.
- Reduced hand calculations for more than 20 hours a week by developing a huge set of reports based on Excel and VBA.

Middle->Senior Data Analyst

Product: Credit card issuing operations processing.

- Successfully led card issuing supporting processes being in charge of 97 team members.
- Reduced the risk of misusing data keeping excellent service level by leading a project of implementing info-security politics on card issuing support process including negotiation with info-security team and developers team.
- Successfully ran an operation line with VBA automation for processing credit card issuing for salary projects capable of processing 2000 records a day by conducting communications with the business department and leading the project with a team of 8 members.This  significantly reduced  time consuming of the business department due to process optimisation. 
- Reduced mistakes rate for 10% and increased speed twice on document processing by leading several VBA-based tools development.

Head of Card Operation Processing (Deputy)

Product: Client operations processing.

Senior [May 2012 - June 2013]:
- Ensured starting a new client service with huge amounts of incoming records keeping the moderate  hand labor amount of a team by developing a VBA-based script for normalising address strings.

Middle [Feb 2011 - Apr 2012]:
- Reduced hand work amount for 95% and helped a team to handle huge amounts of incoming files by developing a VBS- and MS Shell-based tools set for automatisation the process.
- Reduced from 10% to zero the number of late processed documents and ensured excellent service level by developing a VBA-based tools for automated incoming document accounting.

Middle->Senior VBA Developer

Merchant Acquiring Department.
Product: Acquiring service and terminal maintenance.

Senior [Oct 2010 - Jan 2011]:
- Ensured 40 percent of devices in acquiring chain being at service of outsourcing company, correspondingly reducing time consuming with keeping high quality level by leading the project.

Mid-level [Dec 2007 - Sep 2010]:
- Enhanced business planning accuracy up to 90% by developing an Excel-based model using time series techniques for predicting profit amount depending on device count and fees percentage.
- Reduced device repairing time and service unavailability from a week to less than a day avoiding extra parts shortage by developing a VBA based maintenance accounting tool.

Junior [Feb 2006 - Nov 2007]:
- Increased income 5 times enlarging a POS-terminal chain from 50 to 150 contracts and 400+ terminals during a year by providing excellent service and maintenance in a team of 3.

POS-Terminal Engineer

Highly motivated and diligent data scientist with over 20 years of overall experience, including more than  10 years of data-related experience within reputable tech companies. I bring a strong programming background with proficiency in VBA, Excel and Python, extensive knowledge of machine learning libraries and ability of creating models from scratch.

Led an operational process overseeing a team of 97 members. Avid learner with a proven ability to quickly master new technologies, viewing success as the natural result of hard work, continuous skill development, and perseverance. Strong advocate for the power of data-driven approaches combined with creative thinking to optimize processes and deliver impactful results.

I’m seeking a Data Scientist position in a tech company that leverages data-driven strategies daily and drives innovative digital projects, with a mission to improve people’s lives.

Taplio

Anton Bagin's Linkedin Analytics

Anton Bagin

Want to drive more opportunities from LinkedIn?