📋 Description
As a Senior Product Manager for Machine Learning (Biometrics & Document Intelligence), you will own a defined portfolio of computer vision machine-learning capabilities in biometrics or document intelligence, from idea through live production monitoring. Working within ML Product leadership’s standards and priorities, you will connect customer needs, evolving fraud threats, and commercial priorities with ML, data science, and engineering teams, and serve as the named product owner for large banks and financial institutions.
- Strategy: Own capability roadmaps and quarterly priorities aligned with the ML portfolio roadmap and broader product strategy.
- Discovery: Identify customer problems and market opportunities through customer engagement, fraud insights, competitive analysis, and production performance.
- Prioritization: Translate customer, business, and fraud requirements into problem statements, product requirements, and prioritized backlogs; balance new capabilities, model improvements, coverage, and reliability.
- Fraud threats: Incorporate spoofing, deepfakes, injection attacks, and document manipulation into product priorities and evaluation criteria.
- Delivery: Partner with ML engineers, data scientists, engineering, and platform teams on model training, evaluation, deployment, monitoring, and retraining; define feasible deliverables, manage dependencies, and reduce time from model sign-off to live traffic.
- Release: Build in drift detection, feedback loops, rollout criteria, experimentation, and rollback guardrails; challenge unsuitable scope, timelines, or approaches, and recommend readiness with Engineering while documenting accepted risks and rollback plans.
- Performance: Set real-world acceptance criteria covering fraud detection, false acceptance and rejection rates, customer friction, latency, and cost; measure results across segments, geographies, document types, and operating conditions to guide the roadmap.
- Data: Partner with data and labeling operations on data requirements, labeling strategies, and coverage priorities; identify label noise, bias, insufficient data, and coverage-gap risks.
- Fraud defense: Support layered defenses and assess ML capabilities within the broader verification workflow.
- Governance: Own model governance documentation and answer financial-institution customer questionnaires on model risk, audits, and regulation.
- Escalations: Own portfolio escalations, lead root-cause analysis, distinguish verified facts from perceptions, and close with clear actions, owners, and dates.
- Enablement: Give Sales, Customer Success, and Marketing accurate positioning and ensure external performance claims are supported by validated results.
- Document portfolio: Own document-fraud roadmaps and production performance across template and manipulation detection, security feature analysis, classification, synthetic-document detection, image quality, and OCR.
- Capture: Define mobile and web SDK capture requirements, including image quality, corner detection, and auto-capture; measure effects on accuracy and customer conversion.
- Document coverage: Drive global document-type and jurisdiction coverage, including data acquisition and labeling priorities for new and changed documents.
- Extraction: Define accuracy requirements for data extraction and matching, including field-level OCR accuracy and name-matching logic, and close gaps against customer specifications.
- Model service: Own a shared model-scoring and delivery service across platforms, covering APIs, versioning, and integration standards to reduce time from model sign-off to live traffic.
- Document issues: Lead root-cause analysis and customer communication for document escalations, including false rejects, extraction errors, and missed fraud.
- Biometrics: Own biometric roadmaps and production performance across face matching, liveness and presentation attack detection, injection attack detection, deepfake detection, and selfie quality.
- Certification: Lead third-party certification and testing programs such as ISO/IEC 30107-3, CEN/TS 18099, and ETSI TS 119 461, including lab selection, test design, statement of work review, and interpretation for leadership.
- Bias testing: Define and defend demographic bias testing methods, such as FMR and FNMR by demographic group, for bank Fair Lending and model risk reviews.
- Model choices: Benchmark and shadow-deploy third-party and partner models, then recommend whether to build, buy, or partner.
- Attack evaluation: Turn emerging attacks, including deepfakes, virtual cameras, injection tools, and 3D masks, into evaluation, red-team, and roadmap priorities.
- Device changes: Manage performance changes caused by new devices and operating systems through repeatable device data collection and clear rollout criteria.
🎯 Requirements
- Experience: Five or more years in product management, product ownership, or technical delivery roles, including at least three years on ML products in production and demonstrated ownership of priorities and outcomes.
- ML delivery: Experience launching, delivering, or improving production ML products while working directly with data science and ML engineering teams.
- Metrics: Fluency in model performance metrics such as FAR, FRR, APCER, BPCER, ROC curves, and operating points, with statistical judgment about results safe to share externally.
- Technical knowledge: Working knowledge of computer vision and how data quality affects model performance.
- Domain experience: Experience with biometrics such as face matching, liveness detection, and spoof prevention, or document verification, classification, or fraud detection.
- Tradeoffs: Ability to evaluate modeling and data tradeoffs without being a practicing data scientist.
- Customers: Experience working directly with enterprise or bank customers on escalations, technical reviews, or model risk questions.
- Product judgment: Balance customer needs, fraud risk, technical feasibility, commercial value, and operating economics; constructively challenge assumptions and deliver reliable, measurable outcomes.
- Communication: Strong written communication, including PRDs, technical specifications, and one-page executive briefs, and ability to explain technical concepts and limitations to customers and business stakeholders.
- Nice to have: Experience shipping biometric or document identity solutions at scale; experience in identity verification, fraud prevention, financial services, or other regulated environments.
- Nice to have: Familiarity with model monitoring, drift management, adversarial evaluation, and feedback loops; experience with bank model risk frameworks such as SR 11-7 and OSFI E-23, and AI regulation such as the EU AI Act and NIST AI RMF; exposure to pricing and packaging decisions for ML or API-based products.
🏖️ Benefits
- Healthcare: Universal, supplemental, and private healthcare plan choices based on country specifics.
- Financial: Retirement or pension plan contributions and MTK stock plan participation.
- Protection: Life event and disability coverage.
- Time off: Generous annual leave, company holidays, and volunteer time off.
- Learning: E-learning license, tuition reimbursement, and hackathons.
- Home office: Home office setup allowance.
- Optional: Pet insurance, identity theft protection, and legal assistance.