Principal AI - Large Language Models

Lead and manage the design development, and deployment of large language models (LLMs) across our privacy‑first security platform. Responsible for driving end‑to‑end LLM research, engineering, and operational deployment, ensuring model performance, privacy compliance, and real‑world security impact. This role combines technical leadership, hands‑on engineering, and cross‑functional collaboration to integrate LLMs into detection, classification, and response workflows.

Key Responsibilities

• Lead the research, design, and implementation of Transformer‑based architectures (BERT, GPT, T5, LLaMA, etc.) optimized for security tasks.
• Evaluate architectural designs for efficiency, robustness, privacy, and security.
• Define threat‑detection‑specific requirements, including phishing, BEC, malware, and policy violation classification.
• Optimize foundation model pre‑training and distributed training strategies for compute efficiency.
• Develop parameter‑efficient fine‑tuning strategies (LoRA, adapters, prefix tuning) for downstream security tasks.

Small Language Models & Edge Deployment

• Build efficient Small Language Models (SLMs) for edge or customer environments with low latency (<100ms) and memory footprint (<2GB).
• Implement model compression, quantization, pruning, and distillation techniques.
• Ensure robust and secure deployment while maintaining high performance in constrained environments.
• Design and manage dataset pipelines, including privacy‑preserving and synthetic data generation.
• Define evaluation frameworks for adversarial robustness, false positive rates, bias, safety, and interpretability.
• Monitor model performance and implement continuous improvement processes.

Cross‑Functional Collaboration

• Partner with Product and Security teams to translate business requirements into ML solutions.
• Mentor teams and establish best practices for AI safety, privacy, and responsible ML development.
• Align research outcomes with operational deployment and platform capabilities.

Core Technical Expertise

• Transformer architectures and foundation models (BERT, GPT, T5, LLaMA)
• Large‑scale pre‑training, distributed training, and compute optimization
• Parameter‑efficient fine‑tuning (LoRA, adapters, prefix tuning, prompt tuning)
• Model compression and efficiency techniques: quantization, pruning, distillation
• Dataset engineering, synthetic data generation, and privacy‑aware annotation
• LLM evaluation: adversarial testing, bias assessment, robustness, and safety

Qualifications

Education & Experience

• PhD in ML, NLP, CS, or equivalent industry experience (10+ years in ML/AI, with 5+ years focused on LLMs at scale)
• Proven hands‑on experience training, fine‑tuning, and deploying LLMs in production with measurable impact
• Experience in security, privacy, or safety‑critical domains preferred

Preferred Certifications & Achievements

• Publications in top‑tier ML/NLP/security venues
• Experience with edge/mobile deployment or embedded ML systems
• Familiarity with RLHF, alignment, and responsible AI practices

Technical Skills

• Proficient with PyTorch, distributed training frameworks (FSDP, DeepSpeed, Ray)
• Strong understanding of privacy‑preserving ML and security‑focused AI applications
• Experience with inference optimization frameworks (ONNX, TensorRT, vLLM, llama.cpp)
• Scripting for automation and model deployment (Python, Bash, PowerShell)

Soft Skills

• Excellent leadership and team management capabilities
• Ability to make quick, high‑impact decisions under pressure
• Strong communication and presentation skills (technical & non‑technical audiences)
• Analytical thinking, creative problem‑solving, and attention to detail
• Customer‑focused mindset and ability to work in fast‑paced, high‑stakes environments