Offer summary

Qualifications:

Proven experience in data science with a focus on healthcare or operational analytics., Strong programming skills in Python and familiarity with ML/NLP libraries like PyTorch and HuggingFace., Knowledge of statistical and modern deep learning techniques, including transformers., Experience in developing and deploying machine learning models for operational analytics..

Key responsabilities:

Develop and implement LLM and ML-based approaches for chart abstraction from healthcare data.

Design, build, and improve machine learning models to optimize clinical operations.

Create Reinforcement Learning models that utilize physician feedback and patient outcomes.

Contribute to the analytics infrastructure to support decision-making processes in product and business functions.

Job description

Our story

We’re a fast growing, Series A stage healthare AI company building a clinical operating system for MSK practices. We are revenue generating and are looking to carefully expand the team to help us scale to 1,000,000 patients and beyond!

The team is headquartered in NYC with a satellite office in Vancouver, BC. Much of our engineering team is remote - distributed throughout the US and Canada.

We live in Notion docs, Slack channels and Github with weekly meetings over Zoom. Meetings may feature honorary advisors, MedTech execs, and, if we’re lucky, Will’s cat or our Chief Pup Officer.

As Flagler continues to grow, there is a unique opportunity to build the foundations of data and infrastructure to help the product and company reach our full potential. This is where you come in — to design and build reliable, trusted, and timely analytics that accelerate the decision-making process of key product and business functions. You will have a strong impact on the roadmap and growth trajectory of our company.

What you will do

Develop LLM and other ML-based approaches for chart abstraction from both unstructured and structured healthcare data.
Design, build, deploy, and improve machine learning models for operational analytics to help us optimize clinical operations
Design and develop Reinforcement Learning models that leverage physician feedback and patient outcomes to continuously improve clinical performance and AI safety

Required Qualifications

Research, develop, and productionize LLM-powered machine learning methods for extracting insights from longitudinal clinical data (both unstructured and structured).
Design, build, deploy, and improve machine learning models for operational analytics to help us optimize clinical operations
Design and develop Reinforcement Learning models that leverage physician feedback and patient outcomes to continuously improve clinical performance and AI safety
Proven experience in data science, including a proven track record of successful projects in healthcare or operation analytics
Strong programming skills in Python, and fluency with modern data science and ML/NLP libraries (PyTorch, HuggingFace, etc.).
Knowledge of both statistical and modern deep learning (e.g. transformer) techniques and architectures.

Our values

We spend at least as much time with our co-workers as we do with our closest friends + family; if we intend to do the most important and challenging work of our lives, it’s important that these folks energize us, support us, inspire us, and push us to do our best work.

This is what you can expect of your teammates at Flagler:

Persistence + ownership of outcomes: We wear many hats and aren’t afraid to run through walls to solve hard problems.
Personal + professional growth: We push ourselves to learn new things and embrace challenges, even if it means that we sometimes fail.
Don’t take things personally: We value and react quickly to constructive feedback.
Speed is our ally: In the fast-paced world of startups, we understand the value of moving swiftly. We thrive on the adrenaline of working rapidly.
Be Right: We are highly detailed oriented and try to be right, a lot.

Hiring Process

Due to the high volume of applications, we only reach out to candidates selected for interviews. We do not have online leetcode assessments as an initial filter, so we only reach out to very few candidates for initial introduction.