About Spikerz
Spikerz is an AI-powered cybersecurity company defining a new category: protecting enterprises’ external digital assets from account takeover, impersonation, unauthorized access, and harmful content. Global media networks, sports organizations, gaming publishers, and multi-brand consumer companies rely on us, and demand grows as brands run more accounts across more platforms.
About the role
Build the intelligence that identifies emerging threats and powers security investigations. This is an on-site role in Tel Aviv. You’ll own applied AI from research and experimentation through production deployment, combining machine learning, contextual retrieval, and AI agents to deliver measurable protection.
What you’ll do
- Develop detection models that recognize evolving attacks, combining content, account activity, and identity context to identify phishing, impersonation, and account compromise.
- Build AI agents for security investigations, enabling evidence collection, contextual analysis, and response recommendations through tightly scoped tools and explicit authorization.
- Engineer contextual retrieval and security memory, using embeddings, vector search, and reviewed findings to improve investigation accuracy while preserving tenant boundaries.
- Create rigorous evaluation pipelines covering multilingual threats, rare attacks, adversarial inputs, and detection blind spots. Measure precision, recall, and the operational impact of false positives.
- Own the production AI lifecycle: dataset quality, reproducible experiments, model and prompt versioning, controlled releases, drift monitoring, and inference optimization.
What you bring
- 8+ years in applied AI/ML, including production systems whose performance you monitored and improved after deployment.
- Experience at a cybersecurity product company.
- Strong Python and software engineering skills, with practical experience building classifiers, inference services, and data pipelines.
- Hands-on experience with LLMs, structured outputs, embeddings, RAG, vector databases, and tool-calling agents.
- Strong evaluation fundamentals: representative datasets, leakage prevention, imbalanced classification, threshold calibration, and systematic error analysis.
- Experience deploying and operating AI services, including observability, failure recovery, latency, and cost optimization.
Modern AI engineering practices
- Build automated evaluations and adversarial regression tests into the release process.
- Protect against prompt injection, poisoned context, and sensitive-data exposure, and enforce tool permissions independently of model output.
- Use AI coding tools effectively while maintaining independent review, testing, and technical ownership.
Nice to have
- Experience with multilingual NLP, phishing or fraud detection, Milvus, MLflow, Langfuse, Vertex AI, agent orchestration, or MCP integrations.
Why Spikerz?
Turn advances in AI into defenses used by global brands, with direct ownership of detection quality and the next generation of security investigation capabilities.