At Lazarus, we build foundation models to help people solve the world's toughest problems. Enterprise and government clients use our APIs to power products and workflows that keep humans focused on what we do best. We're a venture-backed team headquartered in the Boston, MA area, and we share a passion for innovation, growth, and a commitment to the responsible use of AI.
You can check out a demo of what we do here.
Job Overview
Lazarus is looking for an Artificial Intelligence Research Scientist, who will be at the forefront of designing, developing, and implementing AI models to fundamentally transform the way businesses deal with information stored in all kinds of documents. Your work will contribute to the advancement of our core technologies with an emphasis on addressing the toughest use cases in areas such as document understanding, natural language processing, computer vision, as well as agents that can orchestrate the application of a suite of those tools. This position reports to the Head of AI.
Responsibilities
- Conduct in-depth hands-on research in machine learning, incorporating literature reviews, experimentation, and community conversations as part of a comprehensive strategy to drive the development of novel algorithms and methodologies allowing us to tackle the hardest problems our customers face.
- Design, implement, and iterate on machine learning models to address specific business and technical challenges presented by our customer-facing teams.
- Establish internal processes and expectations for evaluating our models and comparing capabilities to competitive offerings and conveying results to internal and external audiences
- Collaborate with the team to review plans and experimentation results, refine roadmaps based on early data, and develop other team members.
- Stay current with the latest advancements in machine learning research, contributing to the company's thought leadership through publications and presentations and establishing a research roadmap delivering near-term solutions as well as long-term competitive advantage.
- Collaborate with the engineering team to integrate research outputs into production systems, ensuring accuracy, scalability, and reliability.
- Write white papers and blog posts related to the models you construct.
Requirements
- Ph.D. in Computer Science, Artificial Intelligence, or a related field or equivalent experience.
- Previous experience conducting research on, fine-tuning, and building LLMs.
- Mastery of the Transformer architecture, Hugging Face platform, and Python including either PyTorch or Tensorflow.
- Experience delivering deep learning or LLM-based models to industry customers
- Experience with deep learning architectures, reinforcement learning, generative models, or other advanced machine learning paradigms.
- Solid understanding of machine learning fundamentals, optimization techniques, and evaluation metrics.
- Working knowledge of Pandas and NumPY.
- Proficiency in working with large datasets and employing appropriate preprocessing and feature engineering techniques.
- Proven track record of conducting impactful research in machine learning, demonstrated through publications in top-tier conferences or journals.
- Ability to work independently and collaboratively, with excellent problem-solving and communication skills.
- Strong critical thinking and analytical skills, with a passion for staying at the forefront of technological advancements.
Nice-to-haves
- Working knowledge of C++, specifically with CUDA
- Startup experience
- Experience defining and establishing MLOps techniques for enabling Model Lifecycle Development automation
- Document interpretation experience
Benefits
- Comprehensive benefits package, including health, dental, and vision insurance, as well as retirement savings plans
- Flexible working from home arrangements
- Opportunities for growth and professional development
- A collaborative and supportive company culture
- Access to cutting-edge technology and resources for research and development
- Salary range: $150,000-$200,000
Lazarus AI is an equal opportunity employer. We are committed to equal employment opportunity and nondiscrimination for all employees and qualified applicants without regard to a person's race, color, gender, age, religion, national origin, ancestry, disability, veteran status, genetic information, sexual orientation or any characteristic protected under applicable law. We do not tolerate discrimination or harassment of any kind. This applies to every aspect of employment at Lazarus, including, but not limited to, employment, training, promotion, demotion, transfer, leaves of absence and termination.
Top Skills
What We Do
Lazarus applies artificial intelligence and computer vision to develop products and custom solutions to optimize processes. Organizations leverage our APIs to get data out of PDFs to where it needs to be, faster and more accurately than ever before:
RikAI is a large language model that extracts data from any document without training. Regardless of type, format, or language, our advanced document understanding API can contextualize information, analyze layouts, and find answers to natural language questions across any number of documents. Industries from insurance through to legal are using RikAI to use their documents intuitively.
Upload any document and ask almost any question to get the correct answer without training. To improve results, instead of increasing a “training set”, rewrite questions until they consistently yield the best results.
Lazarus Forms is a general data extraction API that makes cluttered, unstructured PDFs as usable as any JSON. Our form extractor pulls entities, tables, and key value pairs from documents without example forms or training. Recognizing both handwritten and typed text, Lazarus Forms extracts more data and requires even less human intervention. We streamline data intake and transfer processes associated with various types of documents; claims forms, new patient intake, attending physician statements, adjuster reports, lab results, ACORDs, and anything that can fit in a PDF.
 (1).png)
