Skip to content
Search jobs inside your AI agent. Claude, ChatGPT, Claude Code, Codex, Gemini, Cursor and Grok
Connect

Mesh Security

NewBe an early applicant

AI / ML Engineer

Posted Updated

Description

We're hiring AI/ML Engineers to design and build these systems — people who combine real machine learning fundamentals with hands-on experience building LLM-powered and agentic applications, and who understand how LLMs actually behave in production. This is a builder role on production AI systems. You'll own pipelines end to end — from model and prompt design through evaluation, cost, latency, and reliability in production.


Responsibilities:

  • Design and implement agentic LLM systems — multi-step reasoning, tool use, and MCP-based integrations — that analyse security data and automate analyst workflows.
  • Build agentic threat hunting: LLM agents that investigate alerts, pivot across the security graph, form and test attack hypotheses, and summarise findings the way a human hunter would
  • Develop user and entity behaviour analytics (UEBA): baseline normal user, identity, and entity activity and detect anomalous behaviour using statistical and ML models
  • Build and improve LLM pipelines for enrichment, classification, and mapping (e. g., tying detections and attack paths to MITRE ATT& CK techniques)
  • Build rigorous evaluation harnesses: measure output quality, catch hallucinations and drift, and make prompt and model changes safe to ship
  • Apply core ML and data science — classification, clustering, anomaly detection, embeddings and retrieval — over large, messy security datasets from dozens of integrations
  • Fine-tune and adapt models where off-the-shelf falls short — for example, tuning embedding models on security data to sharpen retrieval and similarity matching
  • Own production concerns: model selection, cost and latency trade-offs, structured outputs, fallbacks, and monitoring.
  • Work closely with our security researchers to encode their domain knowledge into models, prompts, and evals.


Requirements:

  • 4+ years in machine learning, data science, or AI engineering, with at least one substantial LLM-based system shipped to production.
  • Strong ML and statistics fundamentals — model evaluation, bias/variance, feature engineering, classic supervised and unsupervised methods — not just API orchestration.
  • Hands-on experience building LLM applications: prompt design, structured outputs, RAG and embedding-based retrieval; fine-tuning experience is a plus.
  • Experience designing agentic systems — tool use, planning, multi-agent coordination — with or without frameworks, and a clear-eyed view of their failure modes.
  • Genuine intuition for LLM behaviour: context-window effects, hallucination patterns, prompt sensitivity, and how to design evals that catch regressions.
  • Strong Python and solid software engineering practices — this role ships production code.
  • Prior work on UEBA, fraud detection, or other behavioural anomaly-detection systems.
  • Experience with MCP (Model Context Protocol) or building LLM tool integrations.
  • Exposure to the security domain — detections, vulnerabilities, threat intel, or graph-based security models.
  • Experience with vector databases, streaming systems (Kafka/Redpanda), or AWS-based ML infrastructure.
  • Published work: papers, blog posts, open-source ML/LLM projects, or Kaggle-style competition results.

Skills

Apply

Other locations (2)

See also

ML / AI jobs by country — openings, pay and top skills →