Job description

Senior Machine Learning Engineer, Behavioral Biometrics Remote - United States About the Role Cursive verifies authorship continuously, from first draft to submission. Instead of scanning a finished document and guessing whether a machine wrote it, Cursive analyzes the writing process itself: keystroke dynamics, revision history, and time on task. The result is evidence of how a piece of writing came to be, not a probability score. The privacy model is the hard part and the point. Verification can run via API in the cloud or locally, on the writer's own machine, so the content of what someone writes stays private while they write it. Cursive travels with the writer across Blackboard®, Canvas, Moodle, D2L Brightspace, Google Docs, and Microsoft 365. We enjoy moving fast, shipping new and better features, and delighting our end users. Our engineers are architects, reviewers, and product thinkers: people who know how to ship features end-to-end, leverage any and all advantages from AI, assure the quality of work, and make judgment calls when the requirements are still evolving. On a team this small, there are no handoffs. The person who designs a feature ships it. We're looking for builders who are energized by that scope and opportunity, not overwhelmed by it. Our team owns this space, is first to market , and is driving an industry-wide paradigm shift. Trust in our product, processes, and people are paramount. We are a team of high-integrity individuals who understand that the integrity tool market is an arms race and we intend to win through technical prowess, policy management, and by influencing the nascent narrative focused on human authentication instead of AI detection. You will build the models that decide whether the person typing is the person we enrolled . This is a research role first. You will own problems end to end: framing the question, designing the data collection, building and validating the model, then working with engineering to get it running inside a desktop application with a real latency and memory budget or in the cloud for scaled-server-side processing. We expect academic rigor and we expect the work to ship. What You Will Own • Authorship verification models. This is an open-set verification problem, not classification. You will not have a fixed roster of users, enrollment data is short, and the model has to hold a decision threshold against writers it has never seen. Expect to work on similarity learning, one-class and few-shot approaches, score calibration, and drift within and across sessions. • Features and representations from keystroke telemetry. Extract signal from high-frequency, fine-grained event data: timing distributions, digraph and trigraph latencies, pause and burst structure, editing and revision behavior, effort over time. Decide what earns a place in the model and what is noise . • Inference at the edge. Models run on student laptops alongside everything else the machine is doing. You will optimize for latency, memory, and CPU, and treat quantization and runtime selection as part of the modeling problem rather than something handed off afterward. • Evaluation you can defend. Design subject-disjoint splits, hunt down session leakage, and report false accept and false reject rates rather than accuracy alone. Measure performance across keyboard layouts, device types, non-native typists, and writers with motor differences. Probe resistance to replay and synthetic keystroke generation. When a result looks too good, find out why before anyone else does. • Privacy and fairness by design. These are students, and this is biometric data. You will help decide what never leaves the device, what a fair error rate looks like across populations, and how to explain a verification result to a teacher who has to act on it. • Research into product. Partner with data and software engineers to move models from notebook to release, and stay involved once they are live . Required Skills and Experience • M.S. with 3+ years of related experience, or Ph.D. in Computer Science (or related field). • Direct research experience through graduate lab work, thesis research, or publications. We want to see that you have carried a question from hypothesis through data collection, modeling, and validation, and that you can say clearly what you found and what you did not. • Advanced Python with fluency in the standard stack: scikit-learn, pandas, NumPy, and a deep learning framework such as PyTorch. • Modeling range. Strong with gradient-boosted trees on engineered features ( LightGBM , XGBoost , random forests), and comfortable reaching for sequence models when the temporal structure justifies the cost. • Instinct for what breaks a result. Leakage, distribution shift, small-sample effects, and evaluation setups that flatter the model. • Clear writing. You will produce findings that product, engineering, and institutional customers rely on. • Fluency in written and spoken English. Preferred Skills and Experience • Keystroke dynamics, behavioral biometrics, time-series or telemetry modeling, or text-revision analysis • Open-set verification, metric learning, or one-class methods, including adjacent fields such as speaker verification and writer identification • Model quantization, inference optimization, or deployment to desktop and browser runtimes • Adversarial robustness or presentation attack detection • Education technology, or work under FERPA, GDPR, or biometric privacy law About Blackboard Blackboard is a leading EdTech company and SaaS provider that delivers the digital environment for transformational teaching and learning. We serve thousands of institutions with the industry's most AI-advanced LMS, Blackboard Ally for accessibility, and institutional effectiveness solutions that put educators and learners at the center. We're an education company that builds technology. Learn more at blackboard.com. The expected salary range for this position is $ 127 ,000 - $ 160 ,000. The range reflects base salary only and does not include additional compensation such as company bonus or benefits. Placement within the pay range will depend on a variety of factors, such as experience, skills, internal parity, and location. Candidates must be legally authorized to work in the country where the role is based at the time of hire and must maintain that authorization for the duration of employment. The company does not provide visa sponsorship or immigration support for this position. This job description is not designed to contain a comprehensive listing of activities, duties, or responsibilities that are required. Nothing in this job description restricts management's right to assign or reassign duties and responsibilities at any time.      Blackboard is an equal employment opportunity/affirmative action employer and considers qualified applicants for employment without regard to race, gender, age, color, religion, national origin, marital status, disability, sexual orientation, gender identity/expression,   protected military/veteran status, or any other legally protected factor.

Requirements

Must-haves

Senior Machine Learning Engineer, Behavioral Biometrics
Remote - United States
On a team this small, there are no handoffs. The person who designs a feature ships it. We're looking for builders who are energized by that scope and opportunity, not overwhelmed by it.
You will build the models that decide whether the person typing is the person we enrolled .
What You Will Own
Research into product. Partner with data and software engineers to move models from notebook to release, and stay involved once they are live .
Required Skills and Experience
M.S. with 3+ years of related experience, or Ph.D. in Computer Science (or related field).
Advanced Python with fluency in the standard stack: scikit-learn, pandas, NumPy, and a deep learning framework such as PyTorch.
Modeling range. Strong with gradient-boosted trees on engineered features ( LightGBM , XGBoost , random forests), and comfortable reaching for sequence models when the temporal structure justifies the cost.
Instinct for what breaks a result. Leakage, distribution shift, small-sample effects, and evaluation setups that flatter the model.
Clear writing. You will produce findings that product, engineering, and institutional customers rely on.
Fluency in written and spoken English.
Keystroke dynamics, behavioral biometrics, time-series or telemetry modeling, or text-revision analysis
Open-set verification, metric learning, or one-class methods, including adjacent fields such as speaker verification and writer identification
Model quantization, inference optimization, or deployment to desktop and browser runtimes
Adversarial robustness or presentation attack detection
Education technology, or work under FERPA, GDPR, or biometric privacy law

Nice-to-haves

Preferred Skills and Experience

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

machine learningapiaipythonaccessibility

Platform: jobvite
Effort: medium

Benefits

Benefits detected in the listing.

Vision

Pto