Sr. Data Scientist - AI Voice

Lumeris•Massachusetts, MA

1d•$113,800 - $154,525•Remote

About The Position

At Lumeris, we believe that our greatest achievements are made possible by the talent and commitment of our team members. That's why we are actively seeking talented and collaborative individuals who are passionate about making a difference in the healthcare industry. Join us today as we strive to create a system of care that every doctor wants for their own family and become part of a community that values its people and empowers you to make an impact. We're seeking a Senior Data Scientist to build the voice AI and model capabilities behind Tom, our AI-enabled primary care platform. Reporting directly to our VP of AI, you'll split your time between two halves of the same problem: developing and fine-tuning the models that power clinical AI, and building the voice systems that let us test, stress, and trust those models before they ever reach a patient. This is a hands-on, build-first role for someone who has personally trained models and shipped conversational voice systems into production.

Requirements

Bachelor's degree with quantitative major (e.g. statistics, mathematics, economics, and actuarial science) or equivalent.
5+ year of relevant data science, machine learning, or applied AI experience, or the knowledge, skills, and abilities to succeed in the role.
Must-have hands-on experience building and deploying production voice or conversational AI systems — speech and speech-to-speech, working with real audio in live environments, not text-only NLP or chat.
Must-have demonstrated experience training neural networks from scratch and fine-tuning language models, with the ability to explain different training approaches and when each applies.
Proven track record of shipping, scaling, and hardening AI systems in production.
Strong programming skills in Python and fluency with modern deep learning frameworks (PyTorch, TensorFlow).
Working knowledge of real-time audio infrastructure and telephony integration.

Nice To Haves

Master's or PhD in a quantitative or scientific discipline.
Google Cloud Platform (GCP) experience.
Healthcare or other regulated-domain experience, including familiarity with clinical data standards such as FHIR.
Experience building voice models or conversational agents at an AI lab or voice-first platform.

Responsibilities

Build and deploy real-time conversational voice agents, working across the speech stack — speech-to-text, LLM reasoning, and text-to-speech, as well as newer speech-to-speech approaches.
Solve the practical problems that determine whether voice works in the real world: telephony audio fidelity, latency, interruptions and barge-in, background noise, and variable capture conditions.
Develop simulation capabilities that generate and role-play patient conversations at scale, covering the range of real-world conditions and edge cases that manual testing cannot reach.
Make and defend architecture tradeoffs across the voice stack, balancing control, latency, and simplicity.
Design and train deep neural networks from the ground up, moving beyond off-the-shelf and regression-based approaches.
Fine-tune language models on clinical and longitudinal healthcare data for prediction and generation tasks.
Select and apply the right training approach for the problem, and articulate clearly where training, fine-tuning, and prompting each belong.
Build the data pipelines and transformations that make clinical data usable for model development.
Build evaluation pipelines that measure model output for accuracy, consistency, completeness, and omission — including LLM-as-judge approaches.
Red-team AI agents to surface failure modes before they reach production, and translate findings into measurable quality improvements.
Establish the testing discipline and quality signals that clinical reviewers rely on.
Partner with clinical, product, and engineering teams to turn ambiguous problems into defined technical work.
Communicate complex technical concepts credibly to both technical and non-technical audiences, including company leadership.
Mentor junior team members and raise the technical standard across the group.