Software Research Engineer

ApplyMint; New York (NY); On-site; USD 120,000 - 170,000; Full time

Job summary

Synchron Cognition Labs in New York is seeking Research Engineers to collaborate with Research Scientists on AI/ML models that convert neural signals into action. You will refine models to enable daily control of devices like iPhones or smart homes, while prioritizing data efficiency and safety.

Responsibilities include scaling neural decoding, personalizing models across implant recipients, and ensuring low-latency performance as products move toward clinical and consumer use.

Qualifications

• Experience applying neural signals to real-world devices.
• Ability to translate research concepts into consumer products.
• Proficiency in Python, C++, and deep learning frameworks.
• Bachelor's degree in a relevant field or equivalent experience.
• Experience with real-time or embedded inference is a plus.

Responsibilities

• Scale and optimize neural decoding and multimodal models from endovascular BCI data.
• Develop few-shot personalization and cross-recipient transfer learning.
• Own the path from decoder to product with cross-team collaboration.
• Define metrics for latency, stability, and error recovery in daily use.
• Collaborate with clinical and regulatory teams to drive training objectives.

Skills

Neural signals application

Python

C++

Deep learning (PyTorch)

Research-to-product translation

Education

Bachelor's in CS/EE/BME/Neuroscience

Tools

PyTorch

NVIDIA Holoscan

Job description

About the role:

Synchron builds brain-computer interfaces that restore and protect human agency. Our Stentrode implant is delivered through the blood vessels rather than by open brain surgery - a roughly two-hour procedure in our clinical trial, with most participants going home the next day - and it reads motor intent from inside a vessel adjacent to the motor cortex. People living with severe paralysis have used it to text, browse, shop, and control the devices in their homes by thought alone. Chiral family, the brain foundation model we introduced in 2025, learns directly from that neural data.

Synchron Cognition Labs is hiring Research Engineers to work with Research Scientists to enhance the AI/ML models that turn neural signals into action. This is work that sits close to the product: you will take models trained on a handful of implant recipients and make them reliable enough for daily control of an iPhone, an iPad, an Apple Vision Pro, a smart home, or an LLM-assisted conversation with family.

Two constraints define the work and make it unlike consumer ML. Our clinical data is measured in patient-years, not petabytes - so data efficiency is the research problem, not an optimization. And the model output is not a ranked list; it is someone's only means of speaking, so latency, stability, and failure behavior matter as much as accuracy.

What you will do:

• Scale and optimize neural decoding and multimodal models built on endovascular BCI recordings, and fine-tune Chiral family foundation models for specific control and communication tasks.


• Make small clinical datasets go far. Develop few-shot personalization, and transfer across implant recipients; build calibration that survives signal non-stationarity across sessions, months, and years of implant life.


• Own the path from decoder to product. Work with software, clinical, and human-factors teams to integrate models behind native BCI experiences - including Apple's BCI Human Interface Device protocol, smart-home control, and assistive communication - and hold real-time latency and stability inside what a daily-use assistive device demands.


• Define the metrics that reflect lived use. Go beyond offline accuracy to time-to-target, false-activation rate, recovery after error, and effort per selection; instrument the training and evaluation pipeline so those numbers drive model decisions.


• Close the loop with the people who use the device, partnering with clinical research and participant-facing teams to turn observed use into the next training objective.


• Set technical direction for research projects, and translate findings into internal tooling, publications, and evidence that supports our clinical and regulatory programs.

Minimum qualifications:

• Industry or research experience applying neural signals to real-world signal.


• Experience translating research concepts and devices into consumer products.


• Programming proficiency in Python, C++, or a similar language.


• Hands-on experience with a deep learning framework (PyTorch or equivalent) for both training and inference.


• Bachelor's degree in Computer Science, Electrical Engineering, Biomedical Engineering, Neuroscience, or equivalent practical experience.


• In off

Preferred qualifications:

• Computer Science, Electrical Engineering, Biomedical Engineering, Neuroscience, or a related field.


• Experience decoding neural or biomedical time series - intracortical, ECoG, EEG, EMG, or other neurophysiological recordings - or prior BCI/BMI work.


• Experience optimizing models under hardware constraints: limited compute, limited memory, limited power.


• Experience with real-time or embedded inference - streaming pipelines, low-latency serving, on-device deployment, NVIDIA Holoscan or comparable edge platforms.


• Experience taking models into a product or a regulated medical device (IEC 62304, ISO 13485, FDA software as a medical device).


• Experience working with technical teams of researchers and engineers.


• Experience building assistive or accessibility technology in partnership with the people who rely on it, or working alongside clinical trial teams.