AI/ML Lead
Key Responsibility:
- Lead the fine-tuning and domain adaptation of open-source LLMs (e.g., LLaMA 3) using frameworks like vLLM, HuggingFace, DeepSpeed, and PEFT techniques.
- Develop data pipelines to ingest, clean, and structure cybersecurity data, including threat intelligence reports, CVEs, exploits, malware analysis, and configuration files.
- Collaborate with cybersecurity analysts to build taxonomy and structured knowledge representations to embed into LLMs.
- Drive the design and execution of evaluation frameworks specific to cybersecurity tasks (e.g., classification, summarization, anomaly detection).
- Own the lifecycle of model development including training, inference optimization, testing, and deployment.
- Provide technical leadership and mentorship to a team of ML engineers and researchers.
- Stay current with advances in LLM architectures, cybersecurity datasets, and AI-based threat detection.
- Advocate for ethical AI use and model robustness, especially given the sensitive nature of cybersecurity data
Requirements
Required Skills:
- 5+ years of experience in machine learning, with at least 2 years focused on LLM training or fine-tuning.
- Strong experience with vLLM, HuggingFace Transformers, LoRA/QLoRA, and distributed training techniques.
- Proven experience working with cybersecurity data—ideally including MITRE ATT&CK, CVE/NVD databases, YARA rules, Snort/Suricata rules, STIX/TAXII, or malware datasets.
- Proficiency in Python, ML libraries (PyTorch, Transformers), and MLOps practices.
- Familiarity with prompt engineering, RAG (Retrieval-Augmented Generation), and vector stores like FAISS or Weaviate.
- Demonstrated ability to lead projects and collaborate across interdisciplinary teams.
- Excellent problem-solving skills and strong written & verbal communication.
Nice to Have
- Experience deploying models via vLLM in production environments with FastAPI or similar APIs.
- Knowledge of cloud-based ML training (AWS/GCP/Azure) and GPU infrastructure.
- Background in reverse engineering, malware analysis, red teaming, or threat hunting.
- Publications, open-source contributions, or technical blogs in the intersection of AI and cybersecurity.