Reinforcement Learning Software Engineer

  • Permanent contract 
  • Paris-3e-Arrondissement
  • Share on Facebook
  • Share on Twitter
  • Share on LinkedIn


  • Intelligence artificielle / Machine Learning, SaaS / Cloud Services

Reinforcement Learning Software Engineer

  • Permanent contract 
  • Paris-3e-Arrondissement


Destygo’s mission is to automate 80% of customer support and sales, unleashing team’s potentials by handling repetitive tasks for them so they can focus on high value interactions. To achieve that, we’re building conversational assistants with self-learning capabilities, that improve automatically by analyzing conversations with end-users.

Creation, training and management of our conversational AIs are done directly by business teams with no technical background, on our SaaS platform. Assistants are available through chat and voice on websites, in mobile apps, on Facebook Messenger, via text messages, WhatsApp, etc.


Job description

At Destygo we are building self-learning AI. Our bots are learning to fine-tune their understanding of natural language and to give more precise answers on their own, without human intervention. We are in the world of reward estimation and reinforcement learning, tackling real life challenges with industry-grade software engineering and cutting edge research.

You will be working on our systems from model conception to deployment, watching our bots getting better over time as they talk to more and more users, and improving your models from these real-world experiments.

Join us to live the thrill felt by those on the Apollo program. It is too late to build the first rocket to bring humans to the moon, don't miss building the first industry-grade self-learning conversational intelligence!

Your responsibilities will include:

  • Conceiving and evaluating NLP models
  • Deploying those models to production, integrating them in our pipelines and adapting our code base
  • Creating and maintaining data pipelines to train the models and automatically update them
  • Creating and maintaining internal APIs to serve these models with a high service level availability and fast response times

You will also participate in our research meetings where we conceive and challenge our general NLU algorithms before actually implementing them.

Your work in data science will include:

  • Reading bleeding edge papers on Natural Language Processing, Deep Learning and Reinforcement Learning
  • Writing tailored deep neural network architectures for your use cases
  • Conceiving and developing innovating algorithms and pipelines

Our algorithm and back-end APIs are written in Python 3, and our models are using tensorflow, keras and scikit-learn. You will also need to know basic Unix commands to manage the virtual machines hosting your server.

Preferred experience

This position requires a background in both software engineering and machine learning. We're looking for candidates with at least five years experience in software engineering and a few years experience in machine learning.

We don't require you to be familiar with every technology listed above. Instead, we're looking for candidates that demonstrate enthusiasm and pro-activity. You are expected to be organised in your work, write clean and efficient code and always remain attentive to details.

We look for people who don't choose deep neural networks over linear regression if the latter is enough, but who are not afraid of designing an adversarial learning pipeline when it clearly is the best solution.

You'll find that members of the Tech team are often in front of whiteboards devising better designs or coaching each other, and we expect you to actively be a part of these discussions.

Bonus points:

  • Experience on working on large python projects, adding features, maintaining them and testing them
  • Experience on deploying projects to production and managing Unix servers hosted in the cloud
  • Experience on writing models using machine learning frameworks such as scikit-learn or tensorflow
  • Participation in programming contests (eg Google Code Jam) or data science contests (eg Kaggle contests)

Meet the Destygo team