Adrian S.

Data Scientist

555 dollar

Freelancer

31 years

Bucharest, ROMANIA

Apache Big Data Data analysis Data Science Deep learning Machine Learning Python Solution Architecture Spark

									My experience
								
												More
											Big Telecom CompanyApril 2018 - Present
											 Experience in setting up a Data Science platform (strategy, roadmap, create & maintain backlog); consensus with stakeholders; define KPIs for platform and associated products; deliver blueprint (documentation)
 Benchmark Python programs with previous programs I wrote in R language (machine learning and deep learning); practical advanced analytics with Python (Scikit-learn, Pandas, numpy, PyCharm); Linear Regression, Logistic Regression, Linear Discriminant Analysis, Classification and Regression Trees, Naive Bayes, K-Nearest Neighbors, Learning Vector Quantization, Support Vector Machines, Tree-Based Methods (Basics, Bagging, Random Forests, Boosting), Resampling Methods (cross-validation,, bootstrap), Non-linear Modeling (Polynomial Regression and Step Functions, Splines, Generalized Additive Models), Unsupervised Learning (K-Means Clustering, Principal Components Analysis); how to approach Linear Model Selection and Regularization (Ridge Regression and the Lasso)
 Models assessment (mainly bias-variance trade-off); use peculiar techniques adapted for specific types of problems –classification, regression- (Confidence Interval, Confusion Matrix, Gain and Lift Chart, Kolmogorov-Smirnov Chart, Chi Square, ROC curve, Gini Coefficient, Root Mean Square Error, L1 version of RSME, Cross Validation, Predictive Power)
 Deep learning (Neural Networks mainly for time series forecast); Multilayer Perceptron, Convolutional Neural Network (CNN), Long Short-Term Memory Neural Network (LSTM), hybrid CNN-LSTM; modeling with TensorFlow
 Practice to improve performance; data: clean, resample, transform, rescale; algorithms: evaluation metric, linear versus non-linear; tuning: optimize parameters, random search, grid search; ensembles: blend model predictions; tasks as regards scalability (storage resources, high availability, distributed systems, fault tolerance, synchronous replication, Hadoop), security (authentication, authorization, audit, user permissions, data lineage)

												More
											SAPJanuary 2016 - April 2018
											 Manage incidents resolution (coordination); collaborate with Subject Matter Experts to identify technical solutions by classes (e.g. cloud infrastructure –HEC, Business Intelligence infrastructure, ERP, HANA, NetWeaver, S4HANA etc.)
 Construct a MIM Knowledge Base; build an incident classification relying on IT architectural classes (decision tree); apply also for DevOps; derive conclusions for reducing outage time (achievement in hypervisor area- DevOps like approach); use assigned monitoring tools (Jira equivalent)
 Investigate according with DevOps techniques (Nagios and Splunk for data collection, XML and JSON files)
 Optimize resolution time by applying Automated Guiding Procedures (resolution trees) with probabilities computed for IT architectural classes (optimally searching solutions + new procedures and standard processes)
 Monitor performance, Data analysis and reports to management
 Collaborate with AWS (Amazon Web Services) for services there located; comparison with AWS analytics

												More
											FreelanceNovember 2013 - December 2015
											 R programs for Data Mining: Tree-based, Regression, Time Series, K-Means, Apriori like, Principal Component Analysis, Survival analysis applicable for Forecasting, Classification, Clustering, Association Detection, and Anomaly Detection, Segmentation etc.; design, development of methods, processes, and systems to consolidate and analyse structured/unstructured data to generate productive insights and solutions
 Use CRAN, tidyverse, dplyr
 Technical preparation of data models; shell scripting; apply appropriate numerical methods (e.g. segmentation)
 Benchmark Natural Language Processing solutions
 Use BI tools and visualization (e.g. R Shiny, plotly, Tableau, QlikView)

												More
											SAS Analytics SRLJanuary 2013 - June 2013
											 Participate to analysis & design phase of a SAS data warehouse for a large commercial bank; contribute to application and the overall IT architecture solution design (use Metadata server, choose scheduling and security solution)
 Use Visual Analytics for Data Discovery
 Planning and executing data conversion activities - ETL (Extraction, Transformation, Loading) in jobs (relying on Data Integration Studio)
 Maintenance for ETL jobs (DIS) for a bank
 Demonstration using Visual Analytics, Enterprise Guide, Enterprise Miner
 Design a technical solution for using SAS Enterprise Miner in conjunction with a specific Data Mart using SEMMA methodology (Customer Relationship Management - the solutions envisaged Profiling and Segmentation, Campaign Management, Profitability Analysis etc.)
 Define choice criteria between SAS SPDS solution (Scalable Performance Data Server) and a classical RDBMS
 Enhance a specific Data Mart (Customer Relationship Management) and in conjunction design peculiar data mining applications based on SAS Enterprise Miner; the solutions relied on decision trees, regression (linear, logistic, generalized), time-series (decomposition, forecast, clustering, classification), clustering (k-means, hierarchical), association rules (a priori), principal component analysis, anomaly detection and envisaged:
 Profiling and Segmentation
 Cross-Sell and Up-Sell
 Acquisition and Retention
 Campaign Management
 Profitability and lifetime value
 Market Basket Analysis
 Upgrade credit scoring with survival analysis and extend to campaign management
 Technical demonstration how to use supportive tools to automatically generate and maintain metadata, to keep inventory of data and jobs, to make impact analysis for data model, to accomplish change tracking and release management, to take advantage of performance statistics (mainly with Data Integration Studio)
 Provide technical analysis how to use Data Quality tools
 Driving test planning for DW

												More
											Erste - Romanian Commercial BankJanuary 2002 - January 2012
											 Team coordinator to establish IT Strategies and IT Roadmaps for Erste-BCR (collaboration with business lines in order to align business and IT strategies leading to fulfill business requirements and take advantage of IT systems evolution); use MSOffice suite –Excel, PowerPoint-, Visio, MSProject
 Define IT Enterprise Architecture management process and approach (in an Erste TOGAF inspired environment and COBIT compliant); describe the enterprise architecture model from multiple dimensions with key features for each of them; use Architecture Development Method (AMD);characterize business architecture, data architecture, application architecture and information architecture; inventory of IT standards. Responsible for delivered architectural documentation
 State IT Architecture principles –standards and methods- and subsequent scoring method (with KPIs) for IT projects relying (also agile) on these principles in order to ensure consistent development of systems & solutions; evaluation of projects’ impact
 Attend to implementation of governance, risk, compliance applications
 Prepare/participate/revise/approve technical architecture of the data warehouse (finally a client data warehouse architecture with dependent data marts and two delivery layers, daily and monthly)
 Check the backup/restore/archive and disaster recovery solutions
 Select data mining algorithms (data analytics) and solutions for segmentation and quantitative analysis
 Outline Data Architecture framework and the way towards Information Architecture
 Identify data modeling tools requirements and select metadata tools: Enterprise Architect (from Sparx Systems) and Power Designer (from Sybase - a SAP company)
 Join together Cognos BI (Report Studio, Framework Manager) to the Data Warehouse solution; carry out modeling tasks (build a model, add business logic to the model, create and configure a package); prepare reports (by assembling data source, a model, stored procedure)

												More
											National Bank of RomaniaJanuary 1999 - January 2002
											Main responsibilities as IT Manager:
 Work out the IT strategy
 Implement applications as requested by business lines
 Recruit appropriate IT personnel
 Decrease operational costs Work performed (except projects described in next section):
 Approved IT strategy
 Optimization of information data flow (re-engineering)
 Decreased running costs (about 25%)

												More
											The Institue for Computers (ITC)September 1989 - January 1999
											Positions:
· project manager, senior research engineer
· research engineer
· engineer Lead some research teams for achieving scientific projects related to usage of artificial intelligence knowledge and methodology for decision support (based on original C++ written code):
 “Information system built with neural networks for modeling and forecasting with applications in economy and finance”
 “Applications of modeling with neural networks”
 “Models for structures with neural networks and their simulation”
 Take part at the design and construction of a 32 bits-minicomputer, VAX compatible, where I achieved the implementation of floating point instructions; as a result of this activity I hold the Patent RO 98369

												More
											The Polytechnic University BucharestJanuary 1990 - January 1999
											 Information transmission theory; decision theory (advanced statistics)
 Algorithms for pattern recognition (data mining)
 Propose and supervise student projects (e.g. pattern classification with a Kohonen network)