AI Security & Process Engineer

About the Team

This role resides within the USDS FUSE Intelligence program, an all-hazards team that develops products and services with action-based outcomes to reduce and identify risk to TikTok USDS.

About the Role

As an AI and Security Process Engineer focused on Agentic Systems on the USDS FUSE Intelligence team, you will be at the forefront of designing, building, and deploying sophisticated AI agents that can reason, plan, and execute complex tasks in support of a security organization defending the data of millions of TikTok users. You will work with Large Language Models (LLMs) and agentic frameworks to build solutions that automate workflows, solve intricate security problems, and interact with digital environments dynamically. This role requires a unique blend of strong software engineering skills, deep LLM knowledge, and a creative vision for building autonomous systems that support complex security and analytic processes.

The ideal candidate will possess the skills to execute the roles and responsibilities described below and would optimally have either experience working in a cyber security operations space or familiarity with systems integration, business process development, or other experience developing AI or automation enabled solutions to complex analytic and process problems.

Responsibilities

• Design and Develop AI Agents: Architect, build, and deploy robust, multi-step AI agents capable of complex reasoning, tool use, and decision-making using frameworks like LangChain, AutoGen, or CrewAI. Develop and integrate agentic systems using LLMs, RAG, and tool-use/function-calling frameworks to automate or accelerate complex tasks for end-users.
• RAG: Deep, practical experience with the end-to-end RAG lifecycle, from designing chunking and hybrid retrieval strategies for domain-specific data to implementing agentic design patterns.
• Workflow and Process Design: Translate user needs into robust, autonomous workflows that can reason, plan, and execute actions.
• Integrate Tools and Data Sources: Equip agents with the ability to interact with external systems by integrating APIs, databases, and other data sources to enable real-world action and information retrieval.
• Implement Agentic Architectures: Develop and implement advanced agentic patterns such as ReAct (Reasoning and Acting), plan-and-execute, and multi-agent collaboration to solve complex, open-ended problems.
• Optimize and Evaluate Performance: Establish rigorous metrics to evaluate agent performance, reliability, and efficiency. Continuously iterate on agent design, prompting strategies, and model choice to improve outcomes.
• LLM Customization: Fine-tune and adapt foundational LLMs to enhance their capabilities for specific agentic tasks and domains. Implement and refine Retrieval-Augmented Generation (RAG) pipelines to provide agents with relevant, up-to-date context. Proven experience building and shipping applications powered by Large Language Models (LLMs).
• Collaboration: Work closely with product managers, data scientists, and other engineers to define requirements, identify opportunities for automation, and integrate agentic solutions into our core products and internal workflows.
• Stay Current: Keep up-to-date with the latest advancements in agentic AI, LLMs, and autonomous systems, and champion the adoption of new techniques and technologies within the team.

Minimum Qualifications

• Bachelor's or Master's degree in Computer Science, AI, Machine Learning, or a related technical field.
• 5+ years of professional experience in a software engineering or machine learning role.
• Demonstrable experience building and deploying solutions using agentic AI frameworks (e.g., LangChain, LlamaIndex, AutoGen) and a strong portfolio of projects to showcase this work.
• Deep understanding of Large Language Models (LLMs), transformer architectures, and modern NLP techniques, including prompt engineering, fine-tuning, and RAG.
• Solid software engineering fundamentals in Python, including experience building and deploying production-level APIs and using common ML libraries (e.g., PyTorch, Hugging Face, Scikit-learn).
• Solid software engineering fundamentals, including experience with APIs, databases, and cloud platforms (AWS, GCP, or Azure).

Preferred Qualifications

• Experience designing and building multi-agent systems where agents collaborate or compete.
• Published research in AI/ML conferences (e.g., NeurIPS, ICML, ICLR).
• Experience with MLOps practices and tools (e.g., Docker, Kubernetes, CI/CD pipelines) for deploying and monitoring AI systems.
• A strong passion for the potential of autonomous agents and a creative mindset for finding novel applications.
• Experience building automated tools in Java, C or PowerShell as well as Experience with SQL or other query languages