Want to know if this job is worth applying to?
Jaroměř, Česko
Onsite
Sage Care is a fast-growing, early-stage healthcare startup founded by exceptional leaders from Apple, Uber, Carbon Health and backed by top-tier venture capital (General Catalyst, Chelsea Clinton). With a strong customer pipeline, Sage Care is transforming healthcare by simplifying care navigation.
Our platform makes it easier for patients to find the right doctor and helps providers focus on those who need them most through harnessing the latest AI innovations.
Building on our successful collaborations with health systems across the U.S., we have expanded internationally to the MENA region. We are now partnering with health systems there to deploy our AI-powered care navigation platform.
Every day, our services match real patient queries, symptoms, and pathologies to providers, sites of care, and urgency. These mappings are highly complex and non-linear, ranging from queries like “back pain” to “doctor for head trauma”.
Today, much of this mapping is hand-tuned and heuristic driven, leveraging some NLP tooling and medical expertise, but limited in scope and expandability.
We are looking for someone to own the knowledge encoding piece of this puzzle, helping us learn from real patient data and medical diagnoses and symptoms to help us build out an encoded representation that can translate real patient requests into actionable results.
This role sits at the intersection of ML/AI research and software engineering. We’re looking for someone who can help build out the core routing engine for our agent. You will work closely with engineers who work on the matching software, medical experts and professionals who can help guide an informed, encoded knowledge representation.
Build and own the core symptom and query routing engine
Build models that can map complex queries and symptoms to urgency classifiers, providers, and specialties.
Build learned decision trees that can infer if there are necessary follow up questions to ask to gain more insight into the patient’s specific query.
Leverage insights and learning from real protocols (e.g. Schmidt-Thompson) as well as other triaging SOP’s.
Work with medical professionals to build generalizable representations of how queries and symptoms can map to body systems, specializations, and restrictions.
Learn from real data
Build self-learning models that can learn and iterate from real user data and diagnoses.
Establish metrics of quality and hill climb on these to improve the model in the long term.
Required
7+ years of ML engineering experience
Experience working with foundational ML models (e.g. learned decision trees, deep learning, and reinforcement learning).
Strong backend engineering skills and systems thinking
Experience working with ambiguous problems and defining solutions from first principles
Experience turning research or novel techniques into testable prototypes.
Nice to Have
Experience with evaluation frameworks and model quality measurement
Experience with medical AI systems
Experience designing human-in-the-loop workflows for machine learning.
Example Things you’ve Done
Built out self-learning decision trees to solve complex problems.
Researched and implemented foundational ML models.
Built out self-learning neural network architectures to solve real-world problems.
Leveraged Reinforcement Learning and Markov Decision Processes to learn optimal policies for online systems.