You're seeing this page as if you were . The main menu is still yours, though. Exit from immersion
Malt Strategy
Ismail ErradiIE

Ismail Erradi

Senior Data Scientist · AI & LLM Engineer

€750/day
Paris, FR
3-7 years
Python
Data science
Cloud computing
SIG
LLM

Average response time: 1 hour

About Ismail

Ismail, lead data science (6+ ans) spécialisé en ML, NLP et données géospatiales.
Maîtrise des architectures LLM, agents et RAG pour la mise en production de
solutions IA à fort impact.

-------

Hi, I'm Ismail, Lead Data Scientist with 7 years of experience working at the crossroads of AI and data. Throughout my journey, I’ve tackled projects in climate resilience, GIS data, NLP, LLMs, and responsible AI across different industries.

Today, I’m looking to work on projects that truly matter, in environmental impact, tech innovation, or anything that uses AI to drive positive change. My mission is simple: turn real needs into smart, practical solutions.

If you're looking for a curious, hands-on data expert who’s deeply committed to making a difference, let’s talk!
Insurance
Software Publishing
Environmental and Sustainable Development
Energy and Electricity
Digital and IT Services

  • Arabic

    Native or bilingual

  • Italian

    Basic

  • Spanish

    Basic

  • French

    Native or bilingual

  • English

    Native or bilingual

Can work on-site
Paris (up to 15km)

Experience

  • Mission indépendante
    AI engineer
    TECH
    January 2025 - Today (1 year and 6 months)
    Paris, France
    Mission 1 : Moteur RAG « Psychologie du management »
    • Contexte : corpus de livres et articles sur les biais cognitifs, le leadership situationnel et les comportements organisationnels.
    • Objectif : permettre à des coachs et managers d’interroger ce corpus en langage naturel.
    • Stack : LangChain, ChromaDB, embeddings OpenAI, GPT-4, reranking sémantique, RAGAS (évaluation), Streamlit (front), LangSmith (traçabilité), FastAPI.
    • Résultat : pipeline évalué via RAGAS (faithfulness 0.91, answer relevancy 0.88) sur un panel de 60 questions métier, interface Streamlit livrée aux utilisateurs finaux.

    Mission 2 : Agent conversationnel agentic RAG
    • Contexte : extension du moteur RAG précédent en agent conversationnel multi-tours, capable de raisonner sur plusieurs échanges et de corriger ses propres erreurs de récupération.
    • Stack : LangGraph, architecture corrective RAG, grading des documents récupérés, réécriture de requête en cas d’échec, mémoire persistante entre sessions, LangSmith (monitoring et tests de non-régression).
    • Résultat : l’agent combine le corpus RAG (mission 1), la recherche web en temps réel et les connaissances du LLM pour des réponses multi-sources contextualisées, taux de pertinence validé à 87% via RAGAS sur un panel de 50 questions métier.

    Mission 3 : Agent de planification bien-être
    • Contexte : agent personnel de suivi de séances de méditation et de pratiques bien-être, conçu pour mémoriser les préférences et s’adapter à l’historique utilisateur.
    • Stack : LangGraph, gestion d’état persistant, memory management, tool calling pour la génération de plannings adaptés, LangSmith (traçabilité des sessions).
    • Résultat : agent opérationnel, suggestions personnalisées selon l’historique, planning généré automatiquement en fonction des disponibilités et objectifs déclarés.
    RAG API OpenAI Streamlit Vector Embeddings Agent IA
  • nexqt
    Lead data science
    SOFTWARE PUBLISHING
    October 2023 - December 2024 (1 year and 2 months)
    Paris, France
    Building an AI-powered urban carbon emissions data platform for
    cities, that uses open data, geolocation, GPS and satellites to create
    near real-time GHG emissions estimations at a city, district, or
    building/street level.
    Technical environment: Xarray, geopandas, Rasterio, Sklearn, Keras,
    AWS, Postgis, Postgres, Docker
    Geographic Information Systems (GIS) GHG Protocol Python Machine learning Cloud AWS
  • Axionable
    Senior data scientist
    CONSULTING AND AUDITS
    October 2018 - October 2023 (5 years)
    Paris, France
    R&D project on climate resilience (Switch) as a Tech Lead on heat
    waves and drought. Activities include: monitoring and analysis of
    open data sources on climate, weather and environment (Copernicus,
    DRIAS, Météo-France, etc. ), building a GCP architecture to automate
    processes from data ingestion to visualization, developing a python
    library to read, process & transform raw geospatial data (e.g.
    shapefile, raster, grib, netcdf), extracting insights from climate data
    and displaying them in Data Studio, code review.
    Client: ADEME & Allianz (ongoing), 9 months

    NLP keyword generation to automate the classification of error logs
    according to business categories.The activities include: exploring the
    scientific literature and proposal of approaches, developing APIs for
    and deployment in production using micro services approach.
    Client: General Motors Canada, 6 months

    Audit of trustworthy AI to evaluate the methodology and
    documentation of the clients AI processes according to the market
    standards (LNE, Labelia) and provide recommandations for
    improvements axes
    Client: Crédit Mutuel Arkéa & Orange Business Services, 4 months

    Flavor recommandation engine based on customer's briefs. The
    activities include: data collection, analysis and cleaning, exploring
    multiple embeddings methods, building end-to-end pipelines (from
    data collection to model training and deployment) and developping an
    API to integrate the model into a web application.
    Client: Givaudan, 2 years

    Single-sign-on solution for french media and online publishers.
    Design of the solution architecture and configuration. Development of
    SSO dashboards, site prototypes & monitoring scripts. Drafting the
    solution's integration kit.
    Client: Geste, 5 months
    SIG Python LLM NLP Cloud computing

Recommendations

Be the first to recommend Ismail

Help this freelancer shine by sharing your experience working together.

These freelancer profiles also match your criteria

AgathaA

Agatha Frydrych

Backend Java Software Engineer

4.7

(3)

2

BaptisteB

Baptiste Duhen

Fullstack developer

4.6

(4)

5

AmedA

Amed Hamou

Senior Lead Developer

4

(2)

7

AudreyA

Audrey Champion

Web developer

4.3

(3)

4

Signup to reveal

Education

  • Computer science engineer
    Ecole Polytechnique de l' Université Paris-Sud
    2018
    Computer science engineer
  • Bachelor of Science in Mathematics
    Université Paris 7 Denis Diderot
    2014
    B.A in computer science and mathematics

Skill set

Cloud AWS
Pandas
xarray
keras
OpenAI
Python Flask
FastAPI
Databases
Streamlit
Data visualisation
Backend Service Development
External API integration
Performance Optimization
Testing and Quality Assurance
Containerization and Virtualization
CI/CD Management
Algorithms and Data Structures
Database Management (e.g., SQL, NoSQL)
Cloud Platforms and Infrastructure
Data Cleaning and Preprocessing
Data integration (ETL, ELT and data warehousing)
Deep Learning
Natural Language Processing (NLP)
Model Training and Evaluation
Predictive Modeling
Recommender Systems
Prompt Engineering
AI and Advanced Analytics
Data Visualization
Exploratory Data Analysis
KPIs and Metric Definition and Monitoring
User Segmentation
GitHub
Cursor
geopandas
Hugging Face
RAG
Claude

Categories