AI Model Engineer

About the role

We are looking for an AI Model Engineer to join our AI Pentest team at Aikido Security. You will evaluate newly released open‑source models and adapt the best ones for our code‑auditing and pentesting workflows, helping developers secure software without slowing down delivery.

Key responsibilities

• Scout and evaluate new open‑source AI models.
• Fine‑tune, distill, quantize, or prune models for Aikido’s security agents.
• Build benchmarks measuring vulnerability discovery, accuracy, speed, and cost.
• Optimize and deploy models on a multi‑GPU infrastructure.
• Collaborate with AI engineers and security researchers to move validated improvements into production.

Required profile

• Strong Python and PyTorch expertise.
• Experience fine‑tuning or adapting open‑source LLMs.
• Familiarity with Hugging Face and inference systems such as vLLM.
• Solid experimental and engineering skills.
• Self‑starter who thrives without rigid structure.
• Background with security agents, CUDA/Triton, MoE models, distillation, quantization, or REAP is a plus.

Required skills

• Python
• PyTorch
• Hugging Face
• vLLM
• CUDA
• Triton
• Mixture‑of‑Experts (MoE) models
• Model distillation
• Quantization
• REAP (or similar model‑compression techniques)

What we offer

• Short decision lines and high ownership.
• Opportunity to shape high‑impact projects and ship fast.
• Top‑tier tech compensation and equity.
• High trust, strong values, zero corporate bureaucracy.
• Flexible working hours in an open, informal atmosphere.