xelys jobs xelys jobs

AI Engineer

Netrolynx AI

midpermanentbackenddata United States 137 days ago via LinkedIn

See how well this job matches your profile

Sign up to get an AI match score and generate a tailored application in seconds.

Get your match score

Tags

PythonPyTorchTensorFlowJAXLarge Language ModelsLLM Post-TrainingRLHFRAGVector DatabasesLangChain

About the role

Role Overview

You will build and deploy language-model-powered systems that enhance Twenty’s mission-critical cybersecurity capabilities for U.S. national security. Work across the full lifecycle of AI development, from dataset curation and model fine-tuning to production deployment of reliable inference and retrieval systems.

Responsibilities

  • Create, maintain, and refine high-quality datasets for training and evaluating cybersecurity-focused AI models.
  • Fine-tune language models (from small specialized models to medium-sized foundation models) aligned to mission requirements.
  • Apply post-training/alignment strategies to improve model performance, reliability, and safety.
  • Develop retrieval-augmented generation (RAG) systems that ground outputs in external knowledge bases.
  • Design, develop, and optimize model serving infrastructure for scalable, secure, efficient production deployment.
  • Build evaluation frameworks and testing harnesses to measure model quality, latency, and regression performance.
  • Integrate AI capabilities into operational workflows and applications using modern orchestration frameworks and APIs.
  • Collaborate with product and engineering teams (and security specialists) to translate operational needs into high-performing AI features.

Requirements

  • 4+ years of professional software development experience supporting ML/AI-enabled applications.
  • Strong Python skills.
  • Extensive deep learning experience with PyTorch, TensorFlow, or JAX.
  • Hands-on experience with LLM post-training techniques, including:
    • continued pre-training
    • supervised fine-tuning (SFT)
    • RLHF
    • alignment methods such as DPO, PPO, or GRPO
  • Experience curating, cleaning, and preprocessing datasets for training and evaluation.
  • Working knowledge of relational, graph, and vector database concepts, including tools such as pgvector, ChromaDB, Pinecone, Milvus, or Weaviate.
  • Experience designing evaluation metrics and testing procedures for LLMs and intelligent agents.
  • Ability to integrate LLM/agent systems using Pydantic-AI, LangChain/LangGraph, or CrewAI.
  • Bachelor’s degree in a relevant field or equivalent practical experience.

Nice-to-Haves

  • Experience operating across both cloud and on-prem environments.
  • Strong focus on security, speed, and accuracy in production AI systems.

About Netrolynx AI

Twenty (Netrolynx AI) is a U.S.-based cybersecurity company focused on industrializing offensive cyber operations to deter adversaries and strengthen national security. Backed by major investors, it builds innovative cyber capabilities and develops AI-powered systems to improve mission-critical defenses and security in the digital age.

Scraped 7/9/2026