AI Engineer, Senior Architect

We are seeking an AI Engineer to join our internal R&D team, driving innovative research and development initiatives across advanced intelligence technologies while upholding rigorous data-protection standards. The AI Engineer's main duties will include building goal-driven agent frameworks that can reason, plan, and act in complex scenarios; conducting systematic experimentation and benchmarking against robustness, privacy, and cost constraints; and rapidly prototyping demonstration pipelines that can transform into secure, production-ready inference systems.

The AI Engineer will also produce detailed architectural documentation, data-flow diagrams, and research reports that capture experimental insights. Translating high-level mission requirements into concrete research targets will be key, and the engineer will collaborate closely with product, cyber-operations, and policy teams. Throughout the process, the engineer will share findings through internal presentations and technical briefings, driving the end-to-end lifecycle of AI models from concept to rigorous evaluation and documentation.

Core Responsibilities

• Custom Model Design & Development – Craft novel neural architectures, fine‑tune LLMs, or train domain‑specific models on classified‑grade data sets.
• LLM Agent Workflow Innovation – Design and prototype multi‑step, tool‑use agent frameworks capable of reasoning, planning, and acting in high‑stakes environments.
• Experimentation & Benchmarking – Systematically evaluate model performance, robustness, and efficiency under IL‑2 → IL‑6 constraints, including adversarial resilience and privacy‑compliant metrics.
• Proof‑of‑Concept Development – Rapidly ship technical prototypes that demonstrate feasibility for a future baseline product (e.g., a secure, agent‑centric inference pipeline).
• Architecture & Design Documentation – Produce detailed schematics, data‑flow diagrams, and research reports that capture experimental insights.
• Cross‑Team Engagement – Translate mission requirements into concrete research goals, informing product, cyber‑operations, and policy teams.
• Dissemination & Knowledge Sharing – Present findings internally, write technical briefings and paper drafts, and maintain a private research repository.
• Continuous Learning – Stay abreast of cutting‑edge developments (new kernel methods, prompt‑engineering, policy‑as‑code for agents) and inject fresh ideas into the R&D org.

Technical Requirements

• Education – Bachelor’s (or higher) in Computer Science, Electrical Engineering, Mathematics, or related field; Master’s/Ph.D. in AI/ML, Data Science, or Security strongly preferred.
• Machine Learning – In‑depth knowledge of transformer, diffusion, multi‑modal, and other modern architectures; experience training or fine‑tuning large‑scale models.
• Data Science & Feature Engineering – Ability to engineer domain‑specific features, preprocess encrypted/sensitive data, and build statistically sound training pipelines.
• LLM Agent Design – Proven experience building goal‑driven agent frameworks and multi‑tool orchestration pipelines.
• Experimentation & Evaluation – Skilled at crafting rigorous protocols, benchmarking on multiple metrics, and analyzing results.
• Python & Scientific Stack – Expert with PyTorch, TensorFlow, Hugging Face Transformers, Ray, Dask, and related libraries.
• Model Lifecycle & Governance – Understanding of model provenance, versioning, and lifecycle policy aligned with regulatory constraints (FedRAMP, DoD IL controls).
• Observability & Profiling – Experience using profiling tools (Nsight, PyTorch Profiler) and cost‑tracking for GPU/LLM workloads.
• Policy & Compliance – Familiarity with RBAC/ABAC principles, secure software development lifecycle, and policy‑as‑code for regulated environments.
• Cross‑Domain Collaboration – Comfortable translating high‑level intelligence or cyber‑operations requirements into workable research questions and prototypes.

Preferred qualifications

• Prior work on custom LLMs for defense or intelligence applications (domain‑specific models, hierarchical prompting).
• Experience with semi‑automatic or fully automatic agent orchestration using Temporal, Cadence, or proprietary research engines.
• Publications on model architecture innovations, agent workflows, or secure inference.
• Familiarity with model context protocols or related standards for advanced prompting.
• Hands‑on experience with CUDA/CUDNN, NVIDIA TensorRT, or other GPU optimisation techniques.