5 days ago
Remote, United StatesIntern
H1B sponsor
Responsibilities
- Develop and deploy machine learning and large language models.
- Build and optimize AI agents for automation and decision-making.
- Optimize model inference speed, storage efficiency, and scalability.
- Develop pipelines and MLOps workflows for model training, evaluation, and deployment.
- Contribute to ML, LLM, and agent evaluation and monitoring platforms.
- Experiment with new ML, LLM, and agent technologies.
Requirements
- Be an active student pursuing a master's or PhD in computer science, data science, statistics, mathematics, linguistics, engineering, or a related quantitative field with a strong data focus.
- Proficiency in Python and familiarity with ML libraries such as TensorFlow, PyTorch, or Scikit-learn.
- Experience with GenAI, LLMs, and agentic frameworks such as LangChain and AutoGen.
- Ability to write clean, efficient, and scalable code and solve problems effectively.
- Strong written and verbal communication skills.
- Experience with cloud platforms, ML development tools, and ML deployment tools.
- Preferred familiarity with Node.js or Java.
- Preferred familiarity with LLM fine-tuning, retrieval-augmented generation, vector databases, LLM optimization, vLLM, Hugging Face, or reinforcement learning techniques.
Benefits
- Part-time, work-study-based summer co-op opportunity for active master's and PhD students.
- Location-flexible work options, including office, home, or hybrid work, subject to location and role restrictions.
- Ancestry is an equal opportunity employer and provides reasonable accommodations for qualified individuals with disabilities.
- Employment offers are contingent on a legally compliant background check.
Categories
About Ancestry
Ancestry builds online family‑history research tools and consumer DNA testing for individuals, monetized through subscriptions, records access, and AncestryDNA kits. Founded in 1983 and headquartered in Draper, Utah, it is privately held by Blackstone. Its platform hosts more than 65 billion historical records, supports over 3 million subscribers, and has a DNA network exceeding 27 million people.