freiberufler Data Science Expert auf freelance.de

Data Science Expert

offline
  • auf Anfrage
  • 80807 München
  • Weltweit
  • es  |  en  |  fr
  • 24.02.2022

Kurzvorstellung

Highly skilled Data Science professional with more than 10 years’ exp delivering DS projects for int. corp., combining a solid research publication record and teaching activity in artificial intelligence, and starting up an own fintech venture

Qualifikationen

  • Data Mining
  • Data Science
  • Datawarehouse / DWH
  • Datenanalyse
  • Maschinelles Lernen
  • NonStop SQL & SQL/MX (TANDEM)
  • Product Owner
  • Python
  • team lead
  • Technische Konzeption
  • TensorFlow

Projekt‐ & Berufserfahrung

Co-Founder (Festanstellung)
Veleta.ai, Munich
12/2019 – offen (4 Jahre, 5 Monate)
Finanzdienstleister
Tätigkeitszeitraum

12/2019 – offen

Tätigkeitsbeschreibung

Advanced Backtesting framework

Description:
Generic platform agnostic framework to allow for advanced trading strategies definition and execution combining multiple data resolution and providing an enhance trading description language
Functionality/Methodology:
• Design of highly efficient and highly adaptable data models to represent trading strategies.
• Financial data sourcing and quality control for futures, commodities, forex, etc from different commercial and open sources
• Brain-2-code: a Natural language like interface to specify trading conditions at different execution times (entry point, stay-in-trade, dynamic trailing, dynamic take profit, stop loss, max trade duration conditions)
• Queuing mechanism to support the parallel execution of back-testing jobs.
• Highly scalable processing engine supporting high dimensional volumes of historical data in fine granular resolutions (minutes to days) [10 Mio data points with 1500+ columns]
• Optimized reporting of the results including machine learning generated optimization hints to aid traders, including trade-level insights.
Technologies
• Data manipulation packages (dplyr, lubridate, zoo in R and Pandas in Python)
• Trading specific packages (TTR, Quantmod, Candlestick, QuantPy, SciPy, Pynance)
• Serialization formats (json, yaml)
• Advanced time series in-memory technologies (XTS, TSeries, pandas and scikit-learn) as well as purpose-specific data bases (Influx DB, Timescale DB)
• Custom visualization and dashboarding (ggplot, rcharts, PowerBI, matplotlib, dash)
• Workload management (Slurm)

High-availability intelligent trading bot

Description: multi-level risk-aware machine learning based stochastic optimization for trading strategies
Functionality/Methodology:
• Bayesian hyperparameter tuning for dynamic execution
• Assemble learning and super-learning based facetted optimization
• Enhanced tailor-made Support-Resistance and channels modelling.
• In-trade prediction for dynamic optimal behavior based on Deep Neural Networks
• Optimization based on different risks profiles (from highly-conservative to highly-aggressive)
• Customization of objective function (cumulative net profit, Sortino rate, Sharpe rate, etc)
• Time Series multi-granular prediction techniques
Technologies
• Bayesian optimization frameworks (Scikit-Optimizer)
• Advanced CNN architecture (TensorFlow/Keras)
• Features optimization packages (Boruta, BorutaPy, etc)
• Ensemble learning packages (Superlearner, Scikit-Learb)
• Advanced time series (Prophet, LSTMs, etc)

High-availability intelligent trading bot

Description: robust micro-services-based strategy execution trading bot
Functionality/Methodology:
• Custom integration with industry leading brokers
• Integration with market standards (MT4/MT5)
• Advanced self-healing runtime to ensure minimum downtime and trading state preserving
• Highly scalability supporting the execution several instances in parallel
• Remote client steering based on instant-messaging custom commands
• Instant reporting and secured panic mode
• Customizable risk management and enhanced collision handling for overall portfolio optimization
Technologies:
• High performance asynchronous messaging (ZeroMQ)
• R2MT (for MetaTrader integration), IB-Insync, etc
• Telegram custom development (SM Manager)
• Watchdog steering for self-healing and keep-alive
• Instance balancing with shared data live-feed

Eingesetzte Qualifikationen

Data Mining, Data Science, Datenanalyse, Maschinelles Lernen, Python, Technische Konzeption

INVITED LECTURER - PROFESSOR
International Universirty of La Rioja, Online
12/2016 – offen (7 Jahre, 5 Monate)
Hochschulen und Forschungseinrichtungen
Tätigkeitszeitraum

12/2016 – offen

Tätigkeitsbeschreibung

Accredited professor by the Spanish minister of Education
NoSQL Data Base technologies (I) & (II)
Description: Introduction to NoSQL DB - MongoDB
• Introduction to NoSQL, ACID vs BASE, SQL vs NoSQL
• Different NoSQL Data Bases and when to use them
• MongoDB
• Graph Data Bases
• Neo4j

Creation of Data Science B. Sc. curriculum: Validated by the Ministry of Education

Eingesetzte Qualifikationen

NonStop SQL & SQL/MX (TANDEM)

LEAD ADVERTISING SCIENTIST (Festanstellung)
TELEFÓNICA GLOBAL, Munich
12/2014 – 12/2019 (5 Jahre, 1 Monat)
Telekommunikation
Tätigkeitszeitraum

12/2014 – 12/2019

Tätigkeitsbeschreibung

Team Lead of cross-country team and product owner

Machine Learning Look-Alike Modeling for advanced targeting
Description: Machine Learning Look-alike modeling segmentation from anonymized data in programmatic media

Functionality/Methodology:
• Design of highly efficient and highly adaptable data models to represent users’ behavior and segments
• Features Selection based on K-NN algorithm
• K-NN/Random Forest/K-Means models training
• >80% Accuracy. >50Mio Advertising Ids estimated
Technologies:
• Data manipulation packages Scikit-learn
• Exploratory Data Analysis (numpy, matplotlib, etc.)
• Outlier detection (Boxplot, IQR Score, etc)

Data Anonymization Platform

Description: Design and launch anonymization platform enabling products monetization previously impacted by GDPR

Functionality/Methodology:
• Privacy by design. Design of highly efficient and highly scalable platform to anonymize opted-in customer data in a GDPR compliance way
• Annual and daily encryption seeds to protect customer data. In memory protection for encryption keys
• More than 100Mio records / hour processed
Technologies:
• In memory protection for encryption keys (MEM06-C – mlock)
• AWS – Tailor made Python API for customer data and Pixel API
• MongoDB & Json/bjson file format

Bulk-messaging fraud detection

Description: Real Time monitor for data analysis and automatic fraud detection in Bulk messaging
Functionality/Methodology:
• Design real time monitor for bulk messaging usage for insight extraction and fraud detection prevention
• Real Time detection of fraud usage of bulk message for SIM deactivation
• 75% fraud reduction
Technologies:
• Graphana dashboard and Kafka
• Machine Learning model training (Decision Trees, Random Forest, XGBoost, etc.)

Eingesetzte Qualifikationen

Data Mining, Data Science, Product Owner, Projektmanagement (IT)

Ausbildung

Computer Sciences
PhD. Computer Science
2014
Granada, Spain
Computer Sciences
MsC. in Soft Computing and Intelligent Systems
2010
Granada, Spain
Computer Sciences
MsC. in Computer engineering
2009
Granada, Spain

Über mich

Highly skilled Data Science professional with more than 10 years’ experience delivering data science projects for international corporations, combining a solid research publication record and teaching activity in artificial intelligence, and starting up an own fintech venture

Weitere Kenntnisse

• Co-founder at Veleta.AI. Successfully develop and launch an automatized trading platform based on AI and ML
• Leading deployment of global/local solutions for data activation through a GDPR compliance anonymization
• Evaluating advertising products affected by GDPR and successfully deploying compliance adaption plans
• Designing and leading AI driven campaign management tools and insights visualization
• Experience running cross-country team to internalize the creation of audiences for advertising and retargeting based on 1st party data and look alike modelling through an in-house data anonymization platform
• Hands-on accomplished programmer, Ph.D., lecturer and cited author in recommender systems, personalization, social media, affinity modelling, non-relational data bases, machine learning and artificial intelligent projects, etc.

Persönliche Daten

Sprache
  • Englisch (Fließend)
  • Spanisch (Muttersprache)
  • Französisch (Grundkenntnisse)
Reisebereitschaft
Weltweit
Arbeitserlaubnis
  • Europäische Union
  • Schweiz
Home-Office
bevorzugt
Profilaufrufe
498
Alter
38
Berufserfahrung
14 Jahre und 3 Monate (seit 01/2010)
Projektleitung
6 Jahre

Kontaktdaten

Nur registrierte PREMIUM-Mitglieder von freelance.de können Kontaktdaten einsehen.

Jetzt Mitglied werden