Founding Machine Learning Engineer, Health Algorithms
San Francisco, CA, US
fulltime
Visa Sponsorship
About the role
About Fort
Fort is building a premium health tracking device that helps people keep their bodies strong and capable for life.
We combine wearable sensing, thoughtful product design, and intelligent software to turn training, recovery, and daily physiology into clear, actionable insight. Our first device is screenless, beautifully designed, and built around the idea that strength, recovery, and physical capability should be measured with the same seriousness as sleep, steps, or heart rate.
About the Role
Fort is hiring a Founding Health Algorithms Engineer to build the algorithms that turn raw wearable sensor data into reliable health and performance insights.
This person will work across physiological signal processing, machine learning, sensor fusion, validation, and edge deployment. They will help Fort understand what can be measured, how confidently it can be measured, and when a metric should or should not be shown to a user.
This is not a research-only role. We are looking for someone who can take messy real-world sensor data, develop robust algorithms, validate them properly, and help ship them into a real consumer product.
What You’ll Own
You will own Fort’s health and biosignal algorithm stack from data collection through production.
You’ll work with PPG, IMU, and other wearable signals to develop algorithms for health, recovery, activity, and training insights. You’ll help define the data we collect, the quality standards we hold ourselves to, and the validation required before a metric becomes part of the product.
A big part of this role is knowing how to debug ambiguity. When something fails, you should be able to reason through whether the issue is physiology, hardware, firmware, labeling, data quality, model behavior, or product interpretation.
Responsibilities
Define data collection requirements for Fort’s health, recovery, activity, and training algorithms.
Build PPG, IMU, and biosignal algorithms from raw data through production deployment.
Develop signal quality metrics across the full sensing chain, from sensor behavior to user-facing output.
Create systems for detecting, rejecting, and correcting motion artifacts in wearable sensor data.
Build algorithms for heart rate, HRV, activity, recovery, fatigue, sleep-adjacent signals, exercise detection, and strength training analytics.
Design validation studies that answer practical product questions, not just academic benchmark questions.
Evaluate algorithm performance across different users, conditions, workouts, motion patterns, sensor placements, and populations.
Define confidence scoring, failure modes, and “do not show metric” behavior for health insights.
Work closely with firmware, mobile, hardware, and product teams to deploy algorithms on constrained devices and mobile environments.
Optimize algorithms for low power, low latency, and efficient edge inference.
Translate algorithm behavior into product decisions that are understandable, trustworthy, and useful to users.
What We’re Looking For
We’re looking for someone who has built real algorithms for real sensor data.
You should be comfortable working with noisy, low-signal, motion-corrupted data and turning it into something reliable. You should understand that the hard part is not only building a model, but knowing when the signal is trustworthy, when it is not, and how that uncertainty should affect the product experience.
You may be a strong fit if you have experience with wearable sensing, physiological signal processing, time-series machine learning, embedded ML, or health algorithm validation.
You do not need to have worked on this exact product before. But you should be excited by the challenge of building a system that connects raw biosignals, human physiology, edge constraints, and user-facing health intelligence.
Relevant Experience
Strong candidates may have experience in areas like:
Wearable sensor algorithms
PPG, IMU, or physiological signal processing
Signal quality and motion artifact handling
Heart rate, HRV, SpO₂, sleep, recovery, fatigue, or activity algorithms
Human activity recognition or exercise detection
Time-series machine learning and sensor fusion
Validation against reference devices
Study design for health, fitness, or physiology products
Model calibration and performance evaluation across populations
Embedded ML, edge inference, or constrained model deployment
Working with firmware, hardware, or mobile teams to ship production algorithms.
Why This Role Matters
Fort’s product will only be as good as the trustworthiness of its metrics.
This role will help define what Fort can measure, how confidently we can measure it, and how those measurements become useful to real people. The goal is not to create another dashboard of abstract scores. The goal is to build a wearable that understands training, recovery, and physical capability well enough to give people clear, reliable guidance.
This is a rare opportunity to build the core health intelligence layer of a new consumer wearable platform from the earliest stage.
