Deep-tech work
Train and serve owned models on NVIDIA compute.
Careers
Build clinical AIthat ships.
We're a small team building owned models and agentic workflows for specialty medicine. If you want your work in exam rooms, not slide decks, talk to us.
AI · Infrastructure · Clinical · Design · GTM
Epic FHIRathenahealthNVIDIA InceptionHL7SOC 2HIPAA
Why here
What you get
Hard problems, real users, and a platform-sized mission.
Reliability craft
Verifiers, evals, and guardrails that matter in medicine.
Real impact
Hours returned to clinicians, measured.
Platform scope
Build the layer, expand across the encounter.
Forward-deployed
Sit with the clinicians you build for.
Ownership
Small team, big surface, real autonomy.
Open roles
Where you fit
Roles are illustrative [PLACEHOLDER]; contact us even if nothing matches exactly.
Founding ML Engineer — Speech
Own the Vocalis ASR stack on Riva/NeMo.
ApplyFounding ML Engineer — Reasoning
Build the clinical reasoning + verifier models on NeMo.
ApplyComputer Vision Engineer
Specialty medical imaging on Clara/MONAI for Visera.
ApplyInfrastructure Engineer
TensorRT-LLM/Triton serving and on-prem NIM.
ApplyForward-Deployed Engineer
Embed with specialty practices to ship the wedge.
ApplyClinical Informaticist
Encode specialty workflows, codes, and payer rules.
Apply
01
A real conversation.
We start with your work and what you want to build, not a quiz.
02
A practical exercise.
A scoped, paid problem close to what you'd actually do here.
03
Meet the team.
Talk to the people you'd work with day to day.
04
Decision, fast.
Rolling, respectful, and quick — like our product reviews.
Team
Life at Scribara [PLACEHOLDER]
0
Disciplines hiring
0
Mission
0
Specialties to learn
0
Work that ships
Benefits
How we take care of the team
Competitive compensation and meaningful equity, health coverage, and the budget to do your best work. Specifics are [PLACEHOLDER] on this illustrative site and shared during the process.
Above all: a team that values clarity, craftsmanship, and shipping — and a product whose success is measured in clinician evenings returned.
Our stack
What you'll work with
Deep-tech tools across the owned-model lifecycle.
NVIDIA compute
DGX/H100/H200 for training; Jetson for edge.
Riva + NeMo
Speech and reasoning model training.
Clara / MONAI
Medical-imaging models.
TensorRT-LLM + Triton
Low-latency serving.
RAPIDS
GPU data pipelines for the flywheel.
TS + Python + Rust
Product, ML, and perf-critical paths.
Proof
What clinicians say
Outcomes are reported by design partners and marked illustrative [PLACEHOLDER].
“The fastest I've gone from idea to something a doctor used the next day.”
“Reliability work here is the main event, not an afterthought.”