ML Researcher

AXION; San Francisco (CA); On-site; USD 100,000 - 150,000; Full time

Benefits offered by this job

On-site in SF

Job summary

AXION is seeking an ML Researcher to define and build machine learning systems that translate neural and sensor signals into reliable, real-time user intent. You will work at the intersection of neuroscience, signal processing, and modern deep learning, building models that perform robustly across sessions and environments, and that can run in production on constrained, real-world systems.

Responsibilities include designing state-of-the-art models for decoding intent, building end-to-end ML

Qualifications

• 5+ years of professional experience in machine learning research and/or applied ML.
• Strong Python skills and deep experience with modern ML tooling (PyTorch or JAX).
• Demonstrated ability to take ML projects end-to-end from idea to production.
• Excellent fundamentals in statistical learning, deep learning, and rigorous evaluation.
• Clear communication and collaborative, low-ego approach in a startup environment.

Responsibilities

• Design and train state-of-the-art models for decoding intent from neural and physiological signals.
• Build end-to-end ML pipelines: dataset design, labeling, preprocessing, augmentation, training and deployment.
• Develop denoising and non-stationary signal modeling to improve robustness across users and hardware.
• Explore transformers, state-space models, diffusion/denoising models for time-series and multimodal learning.
• Own offline and online evaluation frameworks with real-time inference constraints.

Skills

Python

PyTorch

ML research

Experiment tracking

Communication

Education

PhD in CS/ML

Tools

JAX

Distributed training

Transformers

Time-series modeling

Job description

The human brain is the most powerful system on Earth, yet we still interact with computers using our fingers — and for millions with motor impairments, even that isn't an option. AXION is building the bridge between human intent and machine action, starting with the people who need it most.

Our first product, AXION Click, lets people control an iPad using nothing but their eyes and brainwaves.

Why build AXION Click?

• Work on technology that changes lives. Your users aren't power users optimizing a workflow, they're people regaining independence. You'll see the impact of your work firsthand.
• Push the frontier of neuroscience. You'll work at the intersection of neuroscience, hardware, and AI on problems that don't have textbook answers.
• So you don't have to tell people you work on B2B SaaS. Brain-computer interfaces are one of the most important technologies of this century. We're the team proving you don't need surgery to get there.

This isn't just assistive tech, it's superhuman. We are creating the foundation for the next era of human-computer interaction. We are a team of passionate researchers and engineers developing technology that doesn't just solve today's accessibility challenges but transforms how everyone interacts with digital devices.

About the Role

As an ML Researcher, you will help define and build the machine learning systems that translate neural and sensor signals into reliable, real-time user intent. You’ll work at the intersection of neuroscience, signal processing, and modern deep learning, building models that perform robustly across sessions and environments, and that can ultimately run in production on constrained, real-world systems. This role is ideal for someone who can do high-quality research and drive it into shipping product.

What You’ll Do

• Design and train state-of-the-art models for decoding intent from neural and physiological signals (e.g., EEG), including multimodal fusion with eye-tracking and contextual signals
• Build end-to-end ML pipelines: dataset design, labeling strategy, preprocessing, augmentation, training, evaluation, and deployment
• Develop algorithms for denoising, artifact rejection, and non-stationary signal modeling to improve robustness across users, sessions, and hardware conditions
• Explore and implement modern architectures (Transformers, state-space models, diffusion/denoising models, self-supervised learning) for time-series and multimodal learning
• Own offline + online evaluation frameworks, including real-time inference constraints (latency, stability, calibration, drift) and metrics aligned with user experience
• Partner closely with hardware and software engineering to integrate models into the product stack and iterate quickly from user feedback and field data
• Help in setting the technical direction for research at AXION: research roadmap, experiments, baselines, ablations, reproducibility standards, and publication strategy where appropriate

What We’re Looking For

Required:

• 5+ years of professional experience in machine learning research and/or applied ML (industry or research labs)
• Strong Python skills and deep experience with modern ML tooling (e.g., PyTorch or JAX, experiment tracking, distributed training)
• Demonstrated ability to take ML projects end-to-end: idea → experiments → strong results → integration into real systems
• Excellent fundamentals in statistical learning, deep learning, and experimental rigor (ablations, leakage prevention, proper validation)
• Clear communication, strong engineering judgment, and a collaborative, low-ego approach in a fast-moving startup environment

Preferred:

• PhD in Computer Science, Machine Learning, Mathematics, or a closely related field
• Publications at top-tier ML venues (e.g., NeurIPS, ICLR, ICML) or similarly strong track record of research impact
• Experience with one or more of:
• Transformers for time-series / multimodal learning
• State-space models / sequence modeling beyond standard RNNs
• Diffusion models, denoising objectives, or probabilistic generative modeling
• Signal processing, EEG/BCI, physiological sensing, or real-world sensor pipelines
• Practical experience with real-time ML, online calibration/adaptation, and deployment constraints (latency, drift, session variability)
• Prior experience at an early-stage startup or owning ambiguous, high-impact projects independently

You’ll Be a Good Fit if You:

• Care about the end user.
• Thrive with ambiguity.
• Know how to 80/20 effectively.
• Love pushing technical boundaries.
• Enjoy working hard and moving quickly
• Want to build something great and be recognized for it.

Compensation

• Compensation Range: $100K – $150K
• Equity: 0.5% – 1.5%

Location

• We are based in San Francisco and work in-person every day from our office in South Park.