Senior ML Engineer – Email Security

• Train, evaluate, and release spam-detection models: build the corpus, pick the threshold, read the precision/recall report, decide ship or retrain.
• Build the production side of the new engine in Go: the inference service, mail-filter integration, tokenizer, and streaming message parser.
• Fix detection quality end to end, from a customer's phishing report to a rule, corpus fix, or retrain, weighing false-positive cost against detection.
• Keep corpus tooling and training infrastructure running.
• Review and approve classifier, rule, and model changes.
• Answer partner engineers' questions about message scoring.
• Write specifications, handbooks, and post-mortems.
• Month 1: ship a rule-model release independently.
• Month 3: merge and run the first piece of the new engine on staging.
• Month 6: release a model while the principal engineer is away.

• 5+ years of shipping detection or classification systems where false positives have a cost you had to defend: anti-spam, anti-phishing, anti-malware, fraud, abuse.
• You have taken at least one ML classifier to production yourself: corpus construction, threshold selection, error analysis.
• Hands-on transformer training in PyTorch: tokenizers, masked-language pre-training, fine-tuning, ONNX export, CPU inference measurement.
• Linux, shell, and Python on shared servers and a shared GPU.
• Daily usage of AI coding agents and tools.
• English at upper-Intermediate (B2) level or higher, with particularly strong written communication skills for technical documentation, specifications, and code reviews.
• Strong plus: Email internals: SMTP, MIME, transfer encodings, SPF / DKIM / DMARC, how Exim, Postfix, and Dovecot handle mail to a filter.
• Strong plus: Experience in creating embedded models.
• Strong plus: Go or Rust for systems code: parsers, streaming, memory budgets, concurrency.
• Strong plus: rspamd or SpamAssassin internals: rules, composites, Lua plugins, mass-check scoring.
• Strong plus: Multi-million-sample corpora, labelling pipelines, annotator tooling.
• Strong plus: Small models for constrained hardware: quantization, distillation, latency budgets.
• Strong plus: LLM-based labelers or validators for scaling annotation.
• Strong plus: Hosting panels: cPanel, Plesk, DirectAdmin.

• A focus on professional development.
• Interesting and challenging projects.
• Fully remote work with flexible working hours, that allows you to schedule your day and work from any location worldwide.
• Paid 24 days of vacation per year, 10 days of national holidays, and unlimited sick leaves.
• Compensation for private medical insurance.
• Co-working and gym/sports reimbursement.
• Budget for education.
• The opportunity to receive a reward for the most innovative idea that the company can patent.