Austin Werner
AI/ML Lead Architect – Large Language Model (LLM) Development
Location: Flexible (Remote)
Overview
We are a forward-thinking firm embarking on the development of a proprietary, fully owned Large Language Model (LLM) tailored to deliver transformative solutions in Legal Services & Documentation, Healthcare & Medical Analysis, and Business Intelligence & Analytics. The model will be linguistically versatile, supporting Arabic and English content, and architected for deployment in cloud and on-premises environments. Responsibilities
Design and architect advanced transformer-based LLMs tailored for legal, healthcare, and business analytics domains. Lead and mentor a high-caliber team of researchers and engineers through the model development lifecycle. Oversee the preparation and curation of large-scale multilingual (Arabic/English) datasets relevant to target domains. Spearhead fine-tuning and training of LLMs from scratch with a focus on domain specialization. Implement and optimize distributed training frameworks (e.g., DeepSpeed, FairScale, Horovod) for scalable model development. Apply state-of-the-art techniques in attention mechanisms, tokenization, model quantization, pruning, and deployment optimization. Evaluate and iterate on open-source models (e.g., LLaMA 2/3, Mistral, CodeLlama, Alpaca) and adapt them for proprietary needs. Collaborate with product stakeholders to ensure solutions are deployable both on cloud and on-premises environments. Establish best practices for model evaluation, benchmarking, and responsible AI deployment, particularly with sensitive legal and medical data. Document technical designs and processes for knowledge sharing and regulatory compliance. Qualifications
5+ years’ experience in AI/ML research and development with a specialization in modern transformer architectures (e.g., GPT, BERT, T5, LLaMA, Mitsral). Proven expertise in LLM fine-tuning and original model training. Robust experience with distributed training frameworks (DeepSpeed, FairScale, Horovod, or similar). In-depth understanding of attention mechanisms, tokenization, and optimization strategies for large neural models. Hands-on work with major open-source LLMs (LLaMA 2/3, Mistral, CodeLlama, Alpaca, etc.). Experience with model quantization, pruning, and deployment optimization for efficient inference across hardware. Record of domain-specific LLM projects within legal, healthcare/medical, or business analytics/finance sectors. Comfortable working in multilingual environments, especially with Arabic and English datasets/content. Preferred Qualifications
Advanced degree (MSc/PhD) in Computer Science, Machine Learning, Artificial Intelligence, or related fields. Familiarity with data privacy, compliance, and model interpretability in regulated domains. Exposure to multi-modal AI (optional but valued). Seniority level
Mid-Senior level Employment type
Contract Job function
Design, IT Industries
Staffing and Recruiting
#J-18808-Ljbffr
We are a forward-thinking firm embarking on the development of a proprietary, fully owned Large Language Model (LLM) tailored to deliver transformative solutions in Legal Services & Documentation, Healthcare & Medical Analysis, and Business Intelligence & Analytics. The model will be linguistically versatile, supporting Arabic and English content, and architected for deployment in cloud and on-premises environments. Responsibilities
Design and architect advanced transformer-based LLMs tailored for legal, healthcare, and business analytics domains. Lead and mentor a high-caliber team of researchers and engineers through the model development lifecycle. Oversee the preparation and curation of large-scale multilingual (Arabic/English) datasets relevant to target domains. Spearhead fine-tuning and training of LLMs from scratch with a focus on domain specialization. Implement and optimize distributed training frameworks (e.g., DeepSpeed, FairScale, Horovod) for scalable model development. Apply state-of-the-art techniques in attention mechanisms, tokenization, model quantization, pruning, and deployment optimization. Evaluate and iterate on open-source models (e.g., LLaMA 2/3, Mistral, CodeLlama, Alpaca) and adapt them for proprietary needs. Collaborate with product stakeholders to ensure solutions are deployable both on cloud and on-premises environments. Establish best practices for model evaluation, benchmarking, and responsible AI deployment, particularly with sensitive legal and medical data. Document technical designs and processes for knowledge sharing and regulatory compliance. Qualifications
5+ years’ experience in AI/ML research and development with a specialization in modern transformer architectures (e.g., GPT, BERT, T5, LLaMA, Mitsral). Proven expertise in LLM fine-tuning and original model training. Robust experience with distributed training frameworks (DeepSpeed, FairScale, Horovod, or similar). In-depth understanding of attention mechanisms, tokenization, and optimization strategies for large neural models. Hands-on work with major open-source LLMs (LLaMA 2/3, Mistral, CodeLlama, Alpaca, etc.). Experience with model quantization, pruning, and deployment optimization for efficient inference across hardware. Record of domain-specific LLM projects within legal, healthcare/medical, or business analytics/finance sectors. Comfortable working in multilingual environments, especially with Arabic and English datasets/content. Preferred Qualifications
Advanced degree (MSc/PhD) in Computer Science, Machine Learning, Artificial Intelligence, or related fields. Familiarity with data privacy, compliance, and model interpretability in regulated domains. Exposure to multi-modal AI (optional but valued). Seniority level
Mid-Senior level Employment type
Contract Job function
Design, IT Industries
Staffing and Recruiting
#J-18808-Ljbffr