Yasser KOTRSI

EMBD Finance acquired the Untapped Global data platform in March 2026; team, platform and scope carried over without interruption. - Designed and shipped the company's AI agent platform from scratch on top of the existing data pipeline, now running in production. An orchestration layer where specialist LLM agents select and sequence work over a shared execution runtime, with every executable step validated against JSON-Schema contracts. - Built the execution substrate that makes agent output auditable: a single run path producing provenance, quality gates and per-tenant usage metering automatically, so every record is traceable from source to output. - Deployed on AWS Bedrock behind a provider-agnostic model gateway, with Step Functions and Lambda for stateless per-step workers, ECS Fargate and Django/DRF for durable orchestration, and a React/TypeScript operator interface. - Designed the governance model for the skill library, CI schema validation and code ownership gates controlling what the runtime is permitted to execute. - Lead a cross-functional team of four (data analyst, two full-stack engineers, one DevOps) while owning platform architecture and contributing hands-on as data engineer. - Run data assessment interviews with portfolio companies, evaluating data infrastructure maturity to inform investment decisions.

- Architected a generic ETL framework reducing data integration time from weeks to 48 hours, enabling dynamic schema adaptation for any data source across African markets - Lead backend development using FastAPI/Django REST APIs while spearheading 3+ proof-of-concepts for emerging data integration technologies - Conduct technical assessments for portfolio companies, evaluating data infrastructure to guide deals team investment decisions - Design end-to-end scalable data pipelines using Python, AWS (Lambda, Glue, S3), Apache Airflow, and Airbyte with CDC capabilities - Develop and package Python SDKs using Poetry for internal distribution and maintain comprehensive code quality pipelines with mypy, black, flake8, and SonarQube - Create interactive analytics ecosystem using Apache Superset connected to PostgreSQL/Snowflake, developing custom chart plugins published to npm - Use infrastructure provisioning using Terraform, Docker, and Kubernetes for scalable cloud architecture

- Core contributor to GP3 accounting platform development, working on ADL functional language compiler and execution scheduler for centralized financial processing across multiple markets. - Developed Python plugins for ADL compiler to validate code block correctness, ensuring reliability of financial computation logic across the platform. - Contributed to a complex Apache Airflow migration to latest version, creating PostgreSQL and Oracle migration scripts for the Chainmanager scheduling solution. - Engineered custom patches for Python libraries to resolve Oracle backend ORM compatibility issues, as this database backend wasn't originally supported by Airflow. - Active member of ADL language community, contributing to productivity tools for functional consultants and improving developer experience. - Technical focus on compiler design, database migrations, and functional programming paradigms in financial software architecture

AMMAS Solution - SINAPSE Project: - Designed and deployed APIs for asset operations cost tracking with interactive chart visualizations - Built generic cloud run service for automated PDF report generation and distribution - Developed web modules for asset lifecycle data validation and import functionality - Created comprehensive reporting features with charts and tables using Django templates - Led product demonstrations to clients during sprint review sessions BDS Solution - IoT & Cloud Infrastructure: - Enhanced AWS Lambda functions processing sensor data with MQTT broker integration - Deployed and configured AWS IoT Greengrass solution on cloud virtual machines - Led technical debt reduction through comprehensive unit and integration testing - Automated GCP deployment using Terraform CI/CD pipelines for multi-environment releases - Active contributor to Scrum rituals and agile development practices Core focus on cloud-native architecture, IoT data processing, and client-facing solution delivery in environmental technology sector.

• Use Agile methodology work process. • Develop new features of data import and data processing for the TRP solution of the company. • Set up a solution of securing the Python/Django code against the built code reverse engineering before deploying it on the client server using Cython. • Set up an asynchronous task processing solution using Rabbitmq broker and Celery Task queue. • Write documentation for the solutions developed. #Technologies: Python, Django, DRF, Celery, RabbitMQ, React JS, HighCharts JS, AWS, Docker, Jira, Gitlab

• Use Arcgis to visualize satellite images and to explore spatial databases. • Implement a C++ solution to generate image features to identify urban objects in an RGB image. • Research and develop new image classification features to characterize the land cover occupation in the urban areas and in towns on the entire French map using the supervised and the unsupervised machine learning classifiers such as K-Means, Random Forest, Gaussian classifier. • Build a CNN classifier. • Use the generated and selected features to train a CNN classifier in order to derive the classes that have not been revealed by the CNN in the previous research work on the classification of the land occupation. • Adjust the previously developed CNN parameters and its design based on the classification measurements. • Use the fine tuning approach to transfer knowledge from trained models to raw models. • Documentation. #Technologies: C/C++, Python, machine learning, CNN, classification, image processing, GIS, Geo Tiff, jupyter-Notebook.

Developing a data mining platform on structured textual data to classify data based on the textual structure and the meaning of the sentences (with the help of the user): - Code refactoring. - Optimize the data display interface. - Develop textual data analysis functionalities and include it to the interface. - Implement a search engine and include many filters to it. - Document application. - Packaging. * Technologies : NLP, machine learning, classification, Java, Swing, XML.

Customize a Ruby on Rails open source solution and host it on AWS EC2 instance: During this internship I have done those tasks : • Understand the Sharetribe solution. • Participate in the brain storming for the new features. • Customize the search engine filters. • Change the display of the photo gallery for each listing in the marketplace. • Document the solution. #Technologies : Ruby on Rails framework, Active Records, Amazon Web Services EC2, MySQL, Web services REST, UML2.0, Trello.

Create a website to manage the share of press releases : - Analyse customer needs. - Set up the WordPress CMS. - Install, configure the online payment modules, add needed adjustments and test it. - Search engine optimization : --- Optimize website template. --- Create filters on the user's post, comments to optimize content. --- Set up back link manager. Technologies : PHP, HTML, Search engine optimization, WordPress, Plugins, MySQL.