François Paupier

Data Engineer Expert Machine Learning

Moves to Nantes

  • 47.2173
  • -1.55392
  • Indicative rate €550 / day
  • Experience 2-7 years
Propose a project The project will only begin when you accept François's quote.

Confirmed availability

Propose a project The project will only begin when you accept François's quote.

Location and geographical scope

Nantes, France
Can work in your office at
  • Nantes and 50km around


Project length
  • ≤ 1 week
  • ≤ 1 month
  • Between 1-3 months
Business sector
  • Defense & Military
  • Software Publisher
  • Energy & Utilities
  • High Tech
  • Internet of Things
+8 autres
Company size
  • 2 - 10 people
  • 11 - 49 people
  • 50 - 249 people
  • 250 - 999 people
  • 1000 - 4999 people
+1 autres



Stack Overflow

Stack Overflow : amiabl amiabl
  • 779 Reputation
  • 19 Bronze
  • 12 Silver
  • 0 Gold


  • Français

    Native or bilingual

  • Anglais

    Native or bilingual

  • Allemand

    Limited working proficiency

Skills (45)

  • Languages
  • Beginner Intermediate Advanced
  • SQL
    Beginner Intermediate Advanced
  • GO
    Beginner Intermediate Advanced
  • C++
    Beginner Intermediate Advanced

François in a few words

Je vous accompagne dans l’industrialisation de vos composants machine learning (ML) afin de passer du prototype à un composant maintenable en production.

J’interviens pour vous accompagner lors 1) du design et de 2) l’implémentation de votre pipeline ML :

1) Je vous propose une architecture ainsi qu’un design reposant sur et des technologies s’intégrant à votre système existant afin d’intégrer un composant ML dans votre infrastructure.

2) J’implémente le pipeline de données permettant le déploiement de vos modèles en production en portant une attention particulière au monitoring et version controlling.

Exigez de votre code ML les mêmes standards d’intégration continu et de maintenabilité que du code « traditionnel ».

Domaines d’expertises ML par ordre décroissant : Traitement d’images, timeseries forecasting, Natural Language Processing (NLP).

Expériences dans les domaines de l’énergie, de l’IoT (event-driven architecture) et du retail.


Tech for good advocate

- Challenge Data ENS en 2019 pour la prédiction du temps d’intervention des véhicules de la Brigade des Sapeurs-Pompiers de Paris.

- Ignite Talk au TensorFlow World 2019, Santa-Clara, USA.

- Rédacteur d’articles techniques, « How-to », focus sur des logiciels open source tel Apache NiFi (plus de 1K claps)



Schneider Electric

Energy & Utilities

Machine Learning Engineer

Greater Boston Area

March 2019 - September 2020

Designed a high-throughput and highly-available ETL pipeline to process IoT data. Pioneered Schneider's Machine Learning Platform to scale and automate analytics deployment.

- Designed a real-time ingestion pipeline for IoT data, based on the Hadoop ecosystem.

- Increased pipeline maintainability and engineer productivity by modeling the ingest pipeline as a finite state machine. Monitored critical state transition with messages collected on Azure Service Bus, giving visibility to SREs to ensure processing-time SLA compliance.

- Reduce analytics component deployment by implementing a Machine Learning framework to automate data preparation and manage ML models' lifecycle.

Leveraged knowledge in domain and data modeling, Python 3.7, Microsoft Azure (especially its messaging product Azure Service Bus), Hadoop ecosystem for big data processing (especially Apache NiFi and Kafka)


Energy & Utilities

Data Engineer

Paris, France

April 2018 - January 2019

Participated in the development of a new commodity tracking platform and revamped the core algorithm for cargo allocation by implementing a graph theory approach.

- Provided our users with exclusive energy market insights by scaling up the ETL pipeline using Python and PostgreSQL to handle an additional source of data.

- Cut down core service runtime by 70% using graph theory. Pitched critical stakeholders over the benefits of a graph resolution strategy and thoroughly documented the new approach.

- Enhanced engineers' productivity on the core service by writing exhaustive test cases using the pytest framework and executed by our CI/CD tool.

Leveraged knowledge in Python 3.7 programming (especially the networkx and pytest library), AWS (EC2 instances and spot market), PostgreSQL, and version controlling with Git.


Co-founder, technical

Nantes, France

January 2018 - September 2018

Envisioned and brought life to a text-generative web app using convolutional neural networks to conceive original rap music lyrics.

- Implemented an ETL in python to build a high-quality corpus of songs used to train the lyrics generation model.

- Built and trained a word-to-word generation model using TensorFlow. Contributed back to the open-source by releasing source-code and sharing the architecture in a blog post.

- Constructed the most cost-effective architecture to serve the model by running a benchmark using a Go proxy server. Saved 90% of monthly bill by shifting from a bare AWS VM to a Google Cloud Run service.

Post Mortem

Press & Media

Founder, Editor and Host

Nantes, France

September 2020 - Today

Creator of the Post Mortem Podcast -

In Post Mortem, engineers reflect on real-life incidents of IT systems they experienced.

- Reach software engineers from top-tier companies and pitch them to share their experience in the show.

- Run interviews with software engineers, experts in machine learning, cybersecurity, IoT to share engineering best practices.

- Monitor outages of leading software companies to learn from past failures and avoid repeating them.

- Edit episode on Adobe Audition to deliver an engaging narrative.

- Develop an awareness campaign on social media to boost podcast visibility and interest in learning from others' mistakes.


High Tech

Machine Learning Engineer

Nantes, France

September 2020 - Today


High Tech

Business analyst

Hong Kong

April 2017 - August 2017

Joined The Storefront in its global growth phase to automate the sales process increasing the number of accounts manageable per clerk.

- Revamped landlords on-boarding process and divided listings time-to-market by 7 (from 3 weeks to 72 hours) using Salesforce workflow.

- Conceived and deployed an email automation campaign to streamline customer feedback collection - web app in JS hosted on

- Led video conference workshops with C levels executives sharing insights on value added by new processes.

- Incentivized sales team to outreach KPIs with an in-CRM gamification process leading to an all-time high in lead generation.

- Supervised change management and thoroughly documented best practices to ease new staff onboarding.

Leveraged knowledge in Salesforce, essentials of web development (vanilla JS and HTML/CSS), business process modelization, and change management.

CHU de Nantes

Medical field

ERP Consultant

Région de Nantes, France

September 2016 - March 2017

Oversaw the deployment of Cerner Millennium ERP in the psychiatry service of Nantes' hospital.



charter modal image

Success is a team effort

Contribute to this success and the community's professionalism by signing the Freelancer Code of conduct

Sign the code