Ciprian Sandu, Ph.D. Email & Phone Number

Bucharest, Romania

Bucharest, Romania

(Some of the) Clients: Syngenta, Holcim, Philip Morris, IBM, Logic Software.(To avoid confusion, Data Rhythm Solutions is my own company; I invoice either on it or as a contractor)Sales ForecastingCement Manufacturing Sensors Predictions: - the data consisted of signals from sensors from machines involved in cement production. - end-to-end (MLOps) workflow for predictions of sensor values for some of these machines (from the moment the data is available until predictions are deployed, including maintenance work)User Profiling:- Data Analytics- Clustering (for creating user profiles)Object Recognition: - Developing computer vision algorithms (using Deep Learning – Convolutional Neural Networks) to recognize the brand of a (wrist) watch once a photo of it is fed.

Bucharest, Romania

- Data Analytics – Use a combination of machine learning, data analysis techniques, and automotive design knowledge to identify distinguishing traits in vehicle bus communication or signal data that can be used to distinguish between normal operation versus degraded health. - Algorithm Design – Design algorithms that detect degraded health, and that can meet the requirements for large datasets operation including footprint, processing speed for detection, predictions of issues while avoiding false alarms. Develop the parameter set for the algorithms, tailored to different configurations of vehicles from different OEMs - Production Support – Work closely with a team of highly skilled software engineers to ensure the algorithms are suitable for production, and that the algorithms faithfully match the design. - Signal Protocols – Quickly learn standard and custom CAN and signals protocols used in the automotive industry. - Development Process – Follow defined software development process. Perform documentation, design, code, and defect reviews.

Leeds, United Kingdom

Leading the Cross Device ProjectThe project was part of a larger Marketing Attribution Model. Its goal was to identify which sessions (visits on the company’s website) were from the same user.The work involved:- Exploratory Data Analysis- Data Preparation- Machine Learning Algorithms: Decision Tree, Random Forests, Gradient Boosted Trees, KNN, K-Means- Putting into production - Mentoring and managing the Junior Data Scientists- Leading the project and driving forward to completion- Working independently, as well as with the team- Planning and communicating project updates to the manager- Learning new technologies and sourcing the right people for support- Prioritizing the workloads for myself and the team- Researching theories and providing clear, concise recommendations to whether further work is pursued- Translating business requirements in data science requirements- Understanding the wider business and how team can impact itResults:The implementation of the K-Means clustering on top of the Decision Tree / Random Forests / Gradient Boosted Trees and the KNN algorithms and the iteration through thousands of scenarios, the performance increased by more than 10%.Technologies:Python - Jupyter Notebook, Numpy, Pandas, Dataiku, Google Cloud Platform, Google Cloud Datalab, Big Query, Microsoft Office Suite

Bucharest, Romania

Leading the Machine Learning Prediction ProjectMaking predictions based on the data history using MachineLearning (Reinforcement Learning and Deep Learning - Feedforward and Recurrent Neural Networks) techniques.LBP is a team of scientists that do football betting (they are not a betting house: they are the betters; they work with betting houses), all based on statistics, mathematics and a lot of software. I was their only Machine Learning person for three years. I have worked on making predictions regarding the number of goals, corners and the overtime shown by the referee at the end of the match half. For one year I have worked with Reinforcement Learning and for two years with Deep Learning.The work involved:- Data Preparation- Reinforcement Learning and Deep Learning- Leading the project up to the final tests for production - Working independently, as well as with the team- Planning and communicating project updates to the CEO- Learning new technologies and sourcing the right people for support- Translating business requirements in data science requirementsResults:For the overtime prediction, for example, I have obtained results that are comparable to the human level performance without using any information on the referee (which is the most relevant, as the overtime is very subjective and dependent on the referee). When I added information on the referee, the performance increased by more than 10%.Technologies:Python, Tensorflow, Numpy, Pandas, Linux Ubuntu, GitHub, Microsoft Azure, Microsoft Office Suite

Bucharest, Romania

Bucharest, Romania

For the client GDF Suez Energy Romania, Bucharest, RO: As a member of the on-site project team, I had the following responsibilities: - to meet with the client, (representatives from several departments: sales, marketing, back office, front office, etc.); - to help create deliverables (the Kick-of presentation, the AS IS document, the Requirements Traceability Matrix, the steering committee presentations, the software comparison presentation); - to help create additional documents (e,g,: meeting minutes for the AS IS and the TO BE processes, lists with all the customers, with all the deliverables, etc.); - to translate the necessary documents for the Accenture experts abroad (so that they can give us feedback), as well as documents for the customer (such as certain deliverables; e.g.: the software comparison presentation). For the client Societe Generale, Lille, France: - to create use case diagrams using Rational Rose software. - to create presentations indicating the relationship between various elements of the application (e.g.: databases and interfaces)

Doctor Of Philosophy (Ph.D.), Machine Learning - Reinforcement Learning, Control Engineering, Blood Pressure Regulation

Automatic Control applied in Medicine. Control Engineering Blood Pressure Regulation. Machine Learning - Reinforcement Learning

Doctor Of Philosophy (Ph.D.), Machine Learning - Reinforcement Learning, Control Engineering, Blood Pressure Regulation

8 months stage.

Master’S Degree, Intelligent Systems, Control Engineeri, 9.30 Out Of 10.00

Activities and Societies: The Engineers OrchestraViolin player in The Engineers Orchestra.

Bachelor'S Degree, Mechatronics, Robotics, And Automation Engineering

Automatic Control Machine Learning - Reinforcement Learning (Rennes, France) Research on the reachability of states of affine systems.

Engineer’S Degree, Control Engineering, 9.30 Out Of 10.00

Activities and Societies: President of the students members of the Romanian Society of Automation and Technical InformaticsI assured.

Mathematics And Computer Science