Location: Mostly remote, occasionally hybrid - Bethesda, MD.

As an AI/LLM Engineer supporting the National Heart, Lung, and Blood Institute (NHLBI), you will leverage strong Python programming skills and provide LLM fine tuning and Prompt Engineering to enable science at NHLBI. You will translate business needs into technical requirements and architecture.

Responsibilities:

• Provide consultative and implementation services to advance the institute’s use of AI to improve operating efficiency and effectiveness.
• Work directly with scientific and business users to evaluate areas where AI can be leveraged to streamline functions related to scientific review of information.
• Support and expand existing AI solutions (both ML and LLM-based) that are using industry LLM’s and services.
• Help assess the effectiveness of existing and potential AI solutions, and provide insights and implementation to improve those solutions.
• Work with NHLBI staff to improve their use of AI solutions through training and consultation.
• Help incorporate AI more broadly into their portfolio of IT solutions and provide insights into NHLBI’s AI direction and roadmap.
• Work with the Enterprise Architecture team to ensure projects follow required standards and objectives.
• Create design, implementation, and operations documentation related to proposed and implemented solutions.

Education:

• Bachelor of Science degree in Computer Science, Information Technology, or related field, or equivalent experience

Experience:

• 10+ years of software development experience w/ at least 2-3 years working with machine learning applications.
• Strong Python programming skills and familiarity with NLP libraries like HuggingFace, Transformers, spaCy, NLTK or LangChain
• LLM-Fine-tuning and Prompt Engineering: Experience focused on fine-tuning opensource LLM’s (LLaMa, Mistral, GPT) and crafting optimized prompts for inference
• Transformer Architecture Understanding: Solid grasp of transformer-based models, attention mechanisms, tokenization and strategies, language modeling principals.
• Data Preprocessing for Language Tasks: Experience preparing and curating text datasets for tasks like text classification, summarization, Q&A, and generative text.
• Model Deployment and Serving: Familiarity with deploying LLMs via APIs using FastAPI/Flask, Docker and serving them efficiently.
• Cloud and Compute infrastructure: Experience training and running LLMs on GPUs using AWS, Azure or GSP with awareness of memory optimization, and inference speed-ups.​
• Ability to obtain an NIH Public Trust.

Desired qualifications:

• Experience with scientific research environment, preferably at NIH.

Terp Techs LLC is an equal opportunity employer.