As a Principal AI Engineer, you will set the technical direction for how Rapid7 builds, tunes and serves LLMs and agentic systems across its security products. You understand LLMs from the ground up, from transformer internals and training to fine-tuning and inference optimization, and you stay hands-on with the hardest problems. You know where LLMs shine and where they fall short, and you treat them as one tool in the AI toolbox, alongside predictive AI and classical ML, choosing the right approach for each problem. Specifically your focus will be to:
- Define the architecture and technical roadmap for LLM and agentic AI systems across Detection and Response and Exposure and Remediation
- Lead model strategy: when to use frontier or open-source models, and when to fine-tune, distill, or post-train
- Fine-tune, distill and post-train open-source LLMs on security data, from data curation through training, evaluation and deployment
- Set engineering standards for context engineering, inference optimization, and evaluation across teams
- Optimize LLM inference at scale with techniques such as speculative decoding, KV-cache management, continuous batching, quantization and custom kernels, balancing quality, latency and cost
- Establish evals as the release gate for every AI system we ship
- Lead design reviews and resolve the hardest technical trade-offs
- Stay hands-on: prototype and build critical model, training and inference code yourself
- Mentor Staff and Senior engineers and shape product and platform investment
The skills and qualities you'll bring include:
- 12+ years of software or ML engineering experience, including 2+ years building production LLM or agentic AI systems. A PhD in Computer Science, Machine Learning, AI or a related field is preferred, or equivalent depth of expertise demonstrated through industry experience and technical contributions
- A track record of building scalable, production-grade AI systems used across multiple teams or products with a strong history of organization-wide impact through technical leadership, mentoring, and shared platforms and standards
- LLMs from first principles:deep understanding of transformer architecture, attention, tokenization and positional encodings, and how models are trained and served, with the ability to implement core components from scratch
- PyTorch: expert-level, including custom model code, training loops, mixed precision and distributed training (e.g., FSDP, DeepSpeed)
- Fine-tuning and post-training:hands-on full and parameter-efficient fine-tuning (LoRA, QLoRA) of open-weight LLMs, distillation, and preference or RL-based methods (e.g., DPO, GRPO)
- Context engineering: retrieval, memory, and context management for long-running agent workflows
- Inference optimization: deep knowledge of prefill and decode, KV-cache management (e.g., PagedAttention, prefix caching), continuous batching, speculative decoding, FlashAttention and quantization, and serving at scale with vLLM, SGLang or TensorRT-LLM
- Evals: defining evaluation strategy and benchmarks for LLM and agent systems
- Strong predictive AI and classical ML fundamentals, and the judgment to choose between them and LLMs based on accuracy, latency and cost
- Strong distributed systems and Python fundamentals
- Takes ownership of significant technical outcomes across multiple teams, proactively identifying and addressing risks, driving complex challenges to resolution and ensuring solutions deliver against agreed technical and customer objectives
- Strong communicator with the ability to explain technical trade-offs to engineering, product, and leadership audiences
- Builds strong cross functional partnerships across Engineering, Product, and other teams to align on technical direction and deliver shared outcomes
- Connects customer needs and business priorities to technical decisions, evaluating the value, cost and trade-offs of AI solutions to ensure investments deliver measurable customer and business outcomes
- Embody our core values to foster a culture of excellence that drives meaningful impact and collective success
Preferred Skills and Experience:
- Experience pre-training or training LLMs from scratch
- Experience writing GPU kernels (CUDA or Triton) and profiling GPU workloads
- Security domain experience, such as detection engineering, SOC/MDR, or vulnerability management
- Experience running AI workloads on AWS at scale
- Publications, patents, or open-source contributions in AI
We know that the best ideas and solutions come from multi-dimensional teams. That's because these teams reflect a variety of backgrounds and professional experiences. If you are excited about this role and feel your experience can make an impact, please don't be shy - apply today.
#LI-CG1#LI-Remote
