Agentic AI Scientist

Description

At Leidos, you'll contribute to AI solutions that serve critical national and global missions-ranging from defense and intelligence to healthcare, energy, and space exploration. Our work emphasizes Trusted Mission AI: systems that are transparent, ethical, resilient, and accountable. You'll collaborate with multidisciplinary teams to transition AI research into operational environments where accuracy, security, and reliability are non-negotiable. Joining Leidos means applying your expertise to solve some of the most complex and meaningful challenges of our time.

We are looking for a motivated Agentic AI scientist who wants to work on challenging problems in a variety of domains - including enterprise IT, health, defense, intelligence, and energy - to get results that apply and go beyond the state of the art for measurably better outcomes. We apply our knowledge, capabilities, and experience to develop and deploy Trusted Mission AI - AI that deserves to be trusted by system owners, end users, and the public - to be accurate, ethical, reliable, and adaptable. We are looking for a applied researcher that is expert in envisioning, developing, and securing AI agents using generative AI and LLM-based tools to transform and add value to human workflows.

Primary Responsibilities.

The Senior Agentic AI Scientist will be creating AI agents to plan, act, assess, and deliver measurable outcomes. They will be working with a team to design, implement and deploy agentic workflows to both automate and optimize labor-intensive workflows, as well as empowering the human workforce to discover entirely new capabilities. As a member of the Leidos AI Accelerator, they will be tasked at different times with both R&D as well as customer-facing goals, to speed the transition of novel research into impact on contract.

The tasks of the Agentic AI Scientist will include fine-tuning and optimizing generative models, potentially applying domain adaptation with supervised and unsupervised approaches. They will develop with a combination of open source models, agentic tools, large proprietary commercial models, and small language models. They will be developing novel approaches to securing agentic workflows and to evaluating the results for accuracy, performance, and impact. They will be expected to ensure AI systems adhere to ethical guidelines, transparency, and fairness principles.

• Bachelor's degree in Computer Science, Engineering or related field and relevant experience, or a Masters degree with relevant experience
• Practical understanding of Large Language Models (LLMs) and agent frameworks such as LangChain, LangGraph, CrewAI, A2A, MCP, or AutoGen
• Practical hands-on experience with generative AI models including prompt engineering, chain-of-thought reasoning, and Natural Language Processing (NLP) tasks such as entity extraction, summarization, and semantic search
• Skilled in data exploration using Python-based tools (e.g., Pandas, NumPy, Jupyter)
• Self-starter with a high degree of intellectual curiosity
• Proficiency in Python and Machine Learning libraries such as Tensorflow
• Ability to obtain a Secret clearance

• Ability to design and implement tool-using AI agents, including API integration, retrieval-augmented generation (RAG), and memory/context management
• Hands-on experience with AI service integration such as NIMS, Azure OpenAI, Bedrock, GCP Vertex AI
• Familiarity with deployment into virtualized and containerized environments (e.g., VMware, Docker, Kubernetes)
• Hands-on GPU programming experience for ML workloads using CUDA, PyTorch, or TensorFlow, including optimization for performance and efficiency.
• Expertise in designing and implementing safety, guardrails, and bias-mitigation strategies for autonomous agents and multi-agent systems
• Experience developing Agentic AI solutions, including autonomous planning-execution-reflection loops, multi-agent collaboration, and coordination at scale
• Familiarity with evaluation and observability tools for AI agents, such as LangSmith, OpenAI Evals, or custom telemetry systems
• Experience integrating agents with cloud-native workflows, streaming data pipelines, and real-time decision-making environments