Amine Mairech Email & Phone Number

Hungary

Budapest, Hungary

Led the development of a Deep Learning framework to forecast thousands of grouped Financial Key Performance Indicators across different business units at Bosch.● Implemented an internal framework for forecasting, led to a significant decrease of time required to experiment with new architectures and more reusable code base using Python’s OOP & Inheritance.● Incorporated transfer learning & fine tuning, led to x4 faster training & improved forecasting performance.● Worked on the design and implementation of new functionalities within the framework of the agile methodology.

Singapore

One of the first hires on the Machine Learning team. Developed product solutions centred on forecasting currency exchange rates for internal clients. Led Research, productionisation and monitoring of Deep learning products.● Implemented feature pipelines, resulted in increased scalability & reproducibility across currency pairs.● Implemented, trained and deployed High-Accuracy Deep Learning Models, resulted in + 80% forecasting accuracy.● Built processing & training pipelines with Docker,AWS and Airflow, led to 5 times faster Dev to Production time.● Incorporated CI/CD scripts for the pipelines, insured Reproducibility and automated product updates.● Built an online feature store using Feast, Redis, pgSQL & Kafka, resulted in faster data transfer across MicroServices & better synchronization between stream and batch processing.● Built performance monitoring & alerting services using Grafana, resulted in efficient system tracking and intervention.● Established a Backtest framework, led to 20 times less latency in experimentation and better selection process.● Created an RL trading Environment and experimented with Deep Reinforcement Learning for automated Decision-Making, resulted in + 90% accurate trading positions.

United States

Conducted research, Implemented and tested various Unsupervised data denoising techniques on seismic data for the purpose of extracting slow slip events signals used for the early detection of earthquakes.● Implemented Deep AutoEncoders for Unsupervised denoising which reduced the noise ratio by 50%.● Tested a Framework for Temporal causal detection and modelled the effects using a sparsity based approach.

Tunis, Tunisia

Developed a Deep Learning based decision support tool allowing the detection of trading signals on currencies.Analysis of foreign exchange markets, consultation of benchmark reference documents for methods of modelling trading signals and implementation of Algorithms composing a tool having as input exchange rates for a given currency pair and giving as output a quantitative appreciation (Rating) of the exchange rate direction in the next time period (1D, XH, 1H, XMin, 1Min)

Engineer'S Degree, Multidisciplinary Engineering

Goal: • The training of high level engineers, able to master the technology, to make it progress and to lead the big projects Admission: •.

Education record

Goal: • Preparate students for engineering studies and national engneering exam Admission: • Admission to IPEIM is through the High School.

Baccalauréat