Staff Data Scientist New

at  FLYR Labs

OUR VISION

We believe in a world where travel companies can innovate freely, growing and accelerating their business, while delivering the experience travelers want and the change the industry needs.
FLYR is a technology company that unlocks freedom to innovate for the travel industry – eliminating legacy constraints to enable real-time decision making and create the experiences travelers seek. With FLYR, businesses are able to improve revenue performance and modernize the e-commerce experience through accurate forecasting, automation, and analytics.

QUALIFICATIONS

A person who’s likely to succeed and thrive in this role will recognize themselves in some of the following areas.

• You like building models and understanding the math behind them. You will build models and architectures from scratch based on AI research papers and your own knowledge and experience.
• You have built models based on modern Reinforcement Learning and ML techniques. Attention, unsupervised RL, planning, simulation, they all play a role in modern optimal control algorithms. At this level, we expect you to hit the ground running and identify ways of improving our current models for better qualitative behaviors and higher quantitative performance.
• You like writing production-ready software in Python to implement your models. At FLYR we believe that the most efficient way to design a model that will work in production is to let the model designers do the implementation and validation. Python (and relevant libraries like Tensorflow and numpy) will be your best ally in that exercise.
• You are meticulous about the quality of your code. Code not tested is most likely wrong. You care about your code being correctly implemented and understandable. Unit testing, validation pipelines and monitoring pipelines are all helpful tools to make sure your code is behaving as expected, and you are keen on working on them.
• You understand and enjoy the challenges of building algorithmic systems. You have built scalable systems in the past. You can skillfully use SQL for data access and manipulation. You understand the value and limitations of different technologies, such as Google Cloud Storage, VertexAI, and BigQuery, and can collaborate with engineers on making architectural decisions about scalable systems.
• You want to succeed as a team. We hope you enjoy working in a cross-functional team, and understand that everyone from developers and data scientists all the way to customer success play an equally important role in a company.
• You take ownership of your work. The team will rely on your expertise and collaboration to solve our algorithm design and implementation problems. This role is suited to someone who likes to work with autonomy within a team of skillful peers.

ABOUT THE ROLE

The scientific challenges that we are facing at FLYR are immense, bringing modern multivariate parametric and non-parametric modeling techniques to commercial operations research. Our research covers a wide range of topics: from time series analysis to Bayesian statistics, Monte Carlo simulations, reinforcement learning and controlled experimentation.
In this role, you will work on Reinforcement Learning pricing models to improve the performance of FLYR’s revenue management product. Airline revenue management is a complex field with many legacy applications and is in much need of disruption. Successful outcomes will make a significant contribution to the state-of-the-art and empower airlines around the world to better compete. FLYR already boasts of a first-of-its-kind RL pricing product, which has provided positive outcomes against legacy providers in controlled experiments. You would play a critical role in defining incremental as well as transformational improvements to the algorithms underlying this product, and help with further product development necessitating this technology.
The work of the data science team is based on end-to-end ownership of models and reproducibility. Model improvements are driven by demonstrability of their impact on our clients in our sophisticated validation framework that mimics CI/CD for software projects.

A bit more about our data science stack, used across our different data science teams, some of which successful candidates will work with:

• Python, Pandas, Polars, SciPy, Scikit-learn, NumPy
• Tensorflow, Dask, dagster and Gurobi to run our models
• Looker for our internal analytics
• Cloud-native technologies and DevOps practices for experimentation and seamless production operations

Come onboard and lead FLYR to new heights with us!

RESPONSIBILITIES

• Work with product and data science management to develop key results for given product objectives.
• Lead technical efforts to improve the performance of deployed models and propose initiatives to shape the long-term scientific vision.
• Apply state-of-the-art reinforcement learning techniques to pricing problems
• Own and actively contribute to the design and development of internal machine learning, pricing and optimization libraries.
• Establish scalable and efficient practices for data mining, ML model development, validation and inference.
• Stay up-to-date on the most recent advances in AI and machine learning

A person who’s likely to succeed and thrive in this role will recognize themselves in some of the following areas.

• You like building models and understanding the math behind them. You will build models and architectures from scratch based on AI research papers and your own knowledge and experience.
• You have built models based on modern Reinforcement Learning and ML techniques. Attention, unsupervised RL, planning, simulation, they all play a role in modern optimal control algorithms. At this level, we expect you to hit the ground running and identify ways of improving our current models for better qualitative behaviors and higher quantitative performance.
• You like writing production-ready software in Python to implement your models. At FLYR we believe that the most efficient way to design a model that will work in production is to let the model designers do the implementation and validation. Python (and relevant libraries like Tensorflow and numpy) will be your best ally in that exercise.
• You are meticulous about the quality of your code. Code not tested is most likely wrong. You care about your code being correctly implemented and understandable. Unit testing, validation pipelines and monitoring pipelines are all helpful tools to make sure your code is behaving as expected, and you are keen on working on them.
• You understand and enjoy the challenges of building algorithmic systems. You have built scalable systems in the past. You can skillfully use SQL for data access and manipulation. You understand the value and limitations of different technologies, such as Google Cloud Storage, VertexAI, and BigQuery, and can collaborate with engineers on making architectural decisions about scalable systems.
• You want to succeed as a team. We hope you enjoy working in a cross-functional team, and understand that everyone from developers and data scientists all the way to customer success play an equally important role in a company.
• You take ownership of your work. The team will rely on your expertise and collaboration to solve our algorithm design and implementation problems. This role is suited to someone who likes to work with autonomy within a team of skillful peers