· Health profile project: The project consisted in Building a health-data platform in order to centralise the storage of health data and standardise health data concepts across the company products. This involved also security aspects such us integrate Oauth flow to authorise users access to specific resources. Kotlin, Spring-boot, Kubernetes, Docker, Sumo-logic, Honeycomb, Mongodb, Nest js, Typescript· Ada-tests project: Took initiative to join a power team responsible for designing, architecting and developing the backend of a new platform in partnership with​ A lab test kit provider. The platform enables customers to purchase home-tests and receive their lab results via the Ada app (https://ada.com/tests/).We developed the backend with a cloud native approach heavily relying on FAAS architecture and AWS resources. The team had full ownership of the order/results/notifications management services and of the underlying infrastructure (IaC). The project was developed following AWS best practices for cloud development. Adopting this approach enabled us to cut infrastructure costs in comparison with the standard K8s cluster provisioning strategy (pay as you go) and reduce deployment complexity (deployment, monitoring and infrastructure provisioning is owned by the dev teams): Kotlin, AWS Lambda, SQS, Eventbridge, DynamoDb, VPC (private,public,NAT..), IAM, ApiGateway, Cloudwatch, S3, CodeBuild, CodePipeline, AWS CDK (IaC)

· Developed an internal service for tracking open pull requests and monitoring metrics related to team codebase. This helped our team to to improve the development process: Kotlin, AWS lambdas, Honeycomb · Helped redesign and implement the reasoning engine library api as part of addressing technical debt: Kotlin· Lead a project in building and automating the generation of data-exchange models. This helped document and structure medical domain models and make them easier to access and consume from core teams/components: Protobuf, Java11, Jenkins· Automated deployment of testing tools in order to enable medical doctors to test new reasoning engine prototypes by introducing the usage of google kubernetes engine and docker instead of manual deployment. This accelerated deployment time from an average of 2 days to 15 mins: Google Cloud Platform, Kubernetes, Docker· Helped develop CI/CD and test pipeline for the AI engine: Travis, Jenkins, Java 11 · Continuous support in introducing new AI features, fixing faulty behaviours of the AI engine and refactoring old code: Java 11, Maven, gradle

• Ada is a Berlin based startup that provides a new approach to health. We deliver a revolutionary virtual healthcare service that gives patients instant access to doctors and tailored medical information, and helps doctors make the best decisions for their patients. My work as AI Scientist consists in designing and developing the mathematical model of the reasoning algorithm that powers Ada.-Languages and Tools: Kotlin, Java, Python, Maven, Gradle, Git, Docker, Jupyter ecosystem, Travis-CI, Jenkins, Kubernetes, Google Cloud Platform, Protocol buffers

• Master thesis at Sony European Technology Centre (Speech and Sound department)• My work at Sony consisted on implementing and analysing the performance of speaker adaptation (modification of the parameters of a speech recognition system or of the input signal towards the characteristics of the current speaker) algorithms described by Microsoft and Google for Deep Neural Networks based systems. First I developed a SOTA Large Vocabulary Continuous Speech Recognition system as a baseline. Then I implemented the described algorithms and performed several experiments with different hyperparameter settings and different adaptation scenarios. My work provided insights of how adaptation works for DNN based ASR systems and showed that positive adaptation results were mostly an artifact of an algorithmic correction of a wrong choice of adaptation hyperparemeters. -Tools: Kaldi, Bash, Matlab

• Intern and work student at the R&D department• My work consisted in developing an "autofocus" algorithm for the Alexa camera. My role was to develop a test platform to perform a statistical analysis of the performance of several algorithms using 4K Imagery. The result of my work consisted on a GUI platform capable of analysing different algorithms and displaying several metrics to determine their overall performance in different shooting environments.As a working student, I continued enhancing the capabilities of the developed platform, and worked on research concerning face detection techniques and their mathematical complexity for integration matters.- Tools: Matlab, C++

• Working Student• I worked as a part of a consultancy team for our client BMW in the lifestyle department. Our mission consisted in working with the sales, logistic and product management departments to solve certification issues of products intended for the Chinese and US market. My job consisted in collecting data over the different products and developing a centralised certification process to overcome short term and long term issues.- Tools: Excel, Visual Basic

• Working Student• My job consisted in helping a team of consultants to perform a market analysis for a german client willing to invest in agricultural industry. The region of interest was North Africa. My role was to collect data, which is known to be sparse in that region , by contacting main players of the concerned industry and local authorities. The collected data served to value the market and evaluate potential investments from our client.-Tools: Excel, PowerPoint

• Bachelor Thesis in R&D Departement• Kinexon developed a high accuracy device capable of giving precise localisation data. My project consisted of implementing the device on a quadcopter and testing its accuracy compared to a GPS localisation device. My work consisted in finding a suitable quadcopter to implement the device on it, and to develop a software interface in order to enable the communication between both devices. Tests were later on designed and conducted to compare flight capabilities with both GPS receiver and Kinexon device. -Tools: Arduino,Bash,C,Matlab

• Intern in R&D Departement• I was accountable for designing PCB Layouts and improve old design for measurement devices boards. subsequent to that, I tested the circuits through simulation tools-Tools: Eagle, LTSpice

• Intern in Energy Transformation Techniques Departement• I was accountable for designing and developing the electronic card responsible for the electronic circuit safety-Tools: Eagle, LTSpice

• Intern in Mobility Departement