Senior Machine Learning Engineer

About Us:

At Halon, our mission is straightforward, yet powerful and complex: empower businesses through dynamic email operations. In an era where email is still the most important communications channel for business success, we provide fully-secure solutions that power innovation with flexibility, scalability and control.

Our Core Values

Clients first - We want our clients to succeed and we exceed their expectations.

Passion & commitment - We are ambitious, motivated by joy, innovative and we love to win.

Move fast, together - We put words into action and clearly communicate goals and expectations.

Trust created by transparency - We are honest and foster a culture of mutual respect and kindness.

To help shape the next chapter of our email security platform, we are seeking an experienced and highly motivated Software Engineer with Machine Learning knowledge to join our email security division. In this role, you will be at the forefront of designing, building, and deploying sophisticated machine learning models to protect businesses from a wide range of email-based threats. You will own the full lifecycle of ML solutions, from data processing and model training to deployment and maintenance in a scalable, cloud-native environment.

What You´ll Do:

• Design and implement cutting-edge ML models using state-of-the-art technologies, including LLMs, Transformers, Reputation systems to detect and mitigate security threats like phishing, malware, and business email compromise.
• Develop robust services for automated threat triage, content anonymization, and the retrieval of security intelligence from vast datasets.
• Build and manage end-to-end Machine Learning pipelines for training, evaluation, and deployment of models for tasks such as threat forecasting and anomaly detection.
• Transform prototypes into production-ready data and ML applications that meet throughput and latency requirements.
• Work with a variety of models, from neural networks to tree-based algorithms, and manage them effectively using tools like MLflow.
• Champion the transition from manual processes to a fully automated MLOps environment, implementing CI/CD pipelines, testing, alerting, and advanced logging to ensure stability and scalability in production.
• Create and maintain ETLs to process large volumes of data for reporting, model training, and critical decision-making, ensuring data integrity and confidentiality.
• Collaborate closely with data scientists, security engineers, and other stakeholders to deploy and maintain core threat detection models in production.

Qualifications & Skills:

Essential

• Master’s or PhD in Computer Science, Machine Learning, or a related quantitative field, or a Bachelor's degree with equivalent senior-level industry experience.
• Experience contributing to multiple highly impactful machine learning projects with proven results.
• Proficiency in Python with the ability to write clean, well-structured, and maintainable code for data analysis, modeling, and experimentation.
• Hands-on experience in the NLP domain involving training, fine-tuning and productionizing transformer-based models for text classification / text-embeddings, with proven experience in LLMs and generative AI.
• In-depth experience with one or more deep neural network frameworks (e.g. PyTorch, Tensorflow, JAX).
• Experience monitoring and maintaining performance of models over time in production, considering model/data drifts.
• Expertise in MLOps, including the use of tools like MLflow, Docker, and Kubernetes.
• A creative mindset, propensity to care deeply about the impact their team has and to encourage novel ways of critical thinking in their team.

Good to Have

• Prior experience in the email security domain is a significant plus. This includes a deep, practical understanding of building and deploying models for spam and phishing detection (analyzing content, headers, sender reputation, URLs, and attachments), targeted threat protection against sophisticated attacks like spear-phishing and whaling, and anomaly detection in email traffic (identifying deviations in communication patterns or volumes). Familiarity with the specific challenges and nuances of modeling for Business Email Compromise (BEC), including CEO fraud and invoice scams, is highly desirable.
• Conceptual understanding of Graph Neural Networks and experience applying GNNs to solve real world problem statements.
• Experience working on large imbalanced datasets, evaluating and selecting models that work well in production on imbalanced real-world data.
• Experience with Rust is a plus.

We welcome applications from people of all genders, backgrounds, and identities.