Role Overview
The Senior ML Engineer will help develop and deploy advanced machine learning technology to protect email infrastructure against spam, phishing, and other evolving cyber threats. You will work on a production-scale security solution that already filters email across approximately 5,000 hosting servers. The role combines hands-on transformer model training with systems engineering to build lightweight, efficient inference technology that runs on customer CPUs with minimal resource consumption. You will take ownership of the complete detection lifecycle, from dataset preparation and model evaluation to production releases and ongoing quality improvements.
Accountabilities
- Model Training and Deployment: Train, evaluate, and release spam-detection models, including building and maintaining training corpora, selecting decision thresholds, analyzing precision and recall, and determining whether models are ready for production.
- Production ML Engineering: Develop the production components of a new email-filtering engine in Go, including the inference service, mail-filter integrations, tokenization, and streaming message parsing.
- Detection Quality Improvement: Investigate customer-reported spam and phishing incidents, trace detection failures, and implement appropriate fixes through rules, dataset improvements, or model retraining.
- Model Performance Optimization: Balance detection rates against false-positive risks, ensuring legitimate messages are not unnecessarily blocked while malicious content is effectively identified.
- Training Infrastructure and Data Tooling: Maintain the corpus management tools and training infrastructure required to support reliable experimentation and model releases.
- Code and Model Reviews: Review and approve classifier, rule, and model changes to improve quality, share technical knowledge, and ensure releases can proceed without depending on a single engineer.
- Technical Collaboration: Help partner engineers understand classification outcomes, explain model scoring behavior, and provide clear technical guidance when questions arise.
- Documentation and Knowledge Sharing: Write technical specifications, handbooks, merge-request descriptions, and post-mortems that make decisions understandable and support effective collaboration.
- Independent Delivery: Progress toward defined milestones, including independently shipping a rule-model release during the first month, delivering a component of the new engine to staging by month three, and supporting a model release independently by month six.
Benefits
- Fully Remote Work: Work remotely with the flexibility to organize your schedule and work from different locations worldwide, subject to role and hiring arrangements.
- Flexible Working Hours: Autonomy to structure your working day around your productivity and commitments.
- Paid Vacation: 24 days of paid vacation annually.
- Public Holidays: 10 days of national holidays.
- Sick Leave: Unlimited sick leave.
- Healthcare Support: Financial compensation toward private medical insurance.
- Wellness Benefits: Reimbursement for co-working spaces and gym or sports activities.
- Learning and Development: An education budget and a focus on ongoing professional growth.
- Innovation Rewards: The opportunity to receive a reward for developing an innovative idea that can be patented.
- Technical Challenges: Opportunities to work on demanding machine learning, cybersecurity, and production engineering projects with measurable real-world impact.
Key Qualifications
- Production Detection Experience: At least five years of experience delivering detection or classification systems where false positives have meaningful consequences, such as anti-spam, anti-phishing, anti-malware, fraud prevention, or abuse detection.
- End-to-End ML Ownership: Experience independently taking at least one machine learning classifier into production, including corpus construction, threshold selection, error analysis, and release decisions.
- Transformer Training Expertise: Hands-on experience training transformers with PyTorch, including tokenizers, masked-language pre-training, fine-tuning, ONNX export, and CPU inference performance measurement.
- Systems and Development Skills: Strong practical knowledge of Linux, shell scripting, and Python, including working with shared servers and GPU resources.
- AI-Assisted Development: Regular use of AI coding agents and development tools to support engineering productivity.
- Written Communication: English proficiency at B2 level or higher, with particularly strong written communication skills for technical documentation, specifications, and code reviews.
- Ownership and Judgment: The ability to analyze model performance, make evidence-based release decisions, investigate difficult problems, and balance detection effectiveness with the cost of false positives.
Additional Valuable Experience
- Email Protocols and Infrastructure: Knowledge of SMTP, MIME, transfer encodings, SPF, DKIM, and DMARC, as well as how Exim, Postfix, and Dovecot integrate with email filters.
- Embedded Machine Learning: Experience building models designed to operate within constrained environments.
- Systems Programming: Familiarity with Go or Rust, particularly for parsers, streaming systems, concurrency, and memory-constrained applications.
- Spam Filtering Frameworks: Experience with rspamd or SpamAssassin internals, including rules, composite rules, Lua plugins, and mass-check scoring.
- Large-Scale Dataset Management: Experience working with multi-million-sample corpora, labeling pipelines, and annotation tools.
- Model Optimization: Knowledge of quantization, distillation, lightweight model architectures, and latency optimization for resource-constrained hardware.
- AI-Assisted Annotation: Experience using LLM-based labelers or validators to scale data annotation and quality assurance.
- Hosting Platforms: Familiarity with hosting control panels such as cPanel, Plesk, or DirectAdmin.
