Yasser Chouadra

Project: Deployment of bills of lading, customs declarations covering several regions (Africa, Europe).

1. Data Engineering: Structuring and processing customs data

● Analysis of business needs and design of LLM prompts for RAG use cases (Vertex AI, LangChain).

● Extraction and automatic structuring of documents via advanced OCR (Google Document AI)

● Implementation of intelligent recommendations: acceptance, refusal, or reinforced control.

● Integration of RAG to query documentary and regulatory databases (HS codes, international standards).

● Development of a business rules engine and real-time risk scoring (inter-document inconsistencies)

● Concrete case: automatic invoice comparison

Technical Environment: Python, Postgresql, BigQuery, Airflow, GCP, LangChain, Spacy

Project: Speech Analytics solutions. I participated in the implementation and optimization of data pipelines for collecting, transforming, and exploiting information from customer relations and telecom operations.

1. Data Engineering: Structuring and processing customer data

● Secure access to customer data & GCP environment setup.

● Automated ingestion of company data to BigQuery.

● Processing, cleaning, and transformation of transcriptions from CallMiner to optimize their exploitation.

2. Machine Learning, RAG & NLP: Conversational analysis

● Development of rules and extraction models to identify intentions, emotions, and key topics.

● Implementation of a categorization and scoring system for interactions via CallMiner.

● Simulation of emotion and sentiment analysis of SFR customers/agents

● Detection of keywords and recurring themes in exchanges.

● Participation in the design of a Semantic Search module with LangChain

Technical Environment: Python, SQL, BigQuery, Airflow, GCP, CallMiner, LangChain, Spacy, PySpark

At CARIAN SOFTWARE DEVELOPMENT, I worked on the Linked LeveL project as a Data Engineer, where I designed an ETL to centralize data from the web and automate the aggregation of synthetic views on applications. I developed high-performance pipelines to collect, transform, and analyze textual data. A NER model allowed the extraction of relevant blocks from filtered web pages. I also created interactive dashboards on the cloud, with multidimensional graphs illustrating entity relationships. Finally, I developed REST APIs to make the solution accessible and integrable.

1. Data Engineer

● Real-time data integration with Python, Pyspark, Postgres (Files, PostgreSQL, WEB)

● Data collection and ingestion using web scraping with Beautifulsoup, Selenium

● Collection of web pages associated with each candidate using GCP - Custom Search API

● Use of Pyspark to apply transformations on data (cleaning)

2. ML & NLP Engineer

● Analysis and exploration of company datasets using Scikit-Learn

● Translation of text blocks (Autocomplete + Translation) with Helsinki-NLP - HugginFace

● Training and evaluation of the Logistic Regression model

● Training and evaluation of the NER Named Entity Recognition model

● Fine-tuning of the NER model

3. Cloud & MLOps Engineer

● Management / Installation of Gitlab runners on the OVH server

● Deploying developed ML models (RL, NER) on Fast API