Senior Natural Language Processing (NLP) Scientist

Who we are at Embroker

Embroker makes commercial insurance simple. 

Since 2015, our team has worked to bring the insurance industry into the 21st century and beyond. Backed with $150M in funding, Embroker is creating the go-to business insurance for high-growth companies. Our digital-first experience combines the best policies with the right rates that fit customer needs.

Nothing is possible without our team. As part of the Embroker Pack, you will have a direct and enormous impact on daily operations, team interactions, and culture. You’ll build cool things, meet great people, and grow with us. All from the comfort of your couch.

We are helping businesses plan for tomorrow, so they can change the world today. Are you in?

About the Data Science & Data Engineering Group

The company has been incredibly successful in selling its flagship business insurance product to institutional backed startups. Now, the company is expanding this offering to all private companies. The data science and data engineering group mandate is to establish a data-driven culture within Embroker by continually applying mathematical techniques to business problems which have not been thought of in a quantitative fashion. Thus far:

• The data science team’s machine learning based pricing models are expected to improve loss ratios by 50% and increase annual revenue by 24x.
• The data engineering team has built a world-class architecture running Apache Airflow, Terraform, AWS Secrets Manager, EMR; ingesting data using Apify; storing in AWS S3 buckets, and delivering a feature store in Snowflake.

The value of this position

The data science team has been tasked with creating a quoting engine to enable automated underwriting and a data architecture to improve customer experience during activation, acquisition, and application. So, this position will be responsible for using NLP to ingest and summarize text to improve pricing models and targeting accuracy.

What you will own in this role

• Work cross-functionally to independently define problem statements, collect data, build analytical models, and drive solutions
• Collaborate with data scientists and data engineers to adapt and scale NLP solutions for a variety of application domains.
• Design innovative and generalized NLP machine learning solutions to extract or infer critical information from unstructured text or multi-modality data. Examples of such solutions include creation or customization of text embedding, classification, Named Entity Recognition, event extraction, topic modeling, and burst detection.
• Closely collaborate with internal and external stakeholders to gain through understanding of their decision making process to develop explainable NLP solutions to support or augment decision making.
• Design evaluation strategies for measuring both model performance and real world impact.

What experience we think is the right fit 

The applicant can be hired at either the Senior NLP, Lead NLP, or Principal NLP level – depending on competencies gained and experience. Job specific requirements are as follows:

Senior NLP Scientist (5-10 years of experience)

The Senior NLP Scientist is expected to have gained knowledge of NLP through academic or personal projects and have at least five years of working experience in a NLP scientist role.

Core competencies:

• Strong general knowledge of natural language processing.
• Ability to independently develop NLP solutions with rigorous evaluation
• Ability to clearly communicate technical solutions to both technical and business leaders
• Keep up with the latest developments in NLP, understand tradeoffs between classic and modern NLP models, specifically pre-trained language models.
• Proven expertise building and validating NLP models with deep learning as well as traditional NLP methods. 
• Strong understanding of the real-world advantages and drawbacks of deep learning and NLP methods
• Ability to handle ambiguity, priorities, and competing objectives. Experience solving highly technical problems.
• Experience in some of the following areas:

• Text processing and construction of corpora;
• Processing of large text collections with standard NLP tools (e.g., scikit-learn, Spacy) for parsing, entity extraction, topic discovery and classification (such as sentiment analysis), and natural language understanding;
• Tuning hyper-parameters of existing NLP models for domain-specific data sets;
• Computational manipulation and analysis of natural language documents using statistical models;
• Experimenting with a large corpora for developing and testing advanced NLP algorithms. 

• Attention to potential data and model biases and develop solutions to mitigate biases.
• Ability to program in Python or other scripting language, comfort with working in AWS cloud.
• Familiarity with common NLP and ML toolkits such as Stanford CoreNLP, OpenNLP, NLTK, scikit-learn, SpaCy, and Tensorflow/PyTorch/Keras.
• Ability to write clean, understandable code that follows leading industry standards and practices and is well-documented, and to build easily reproducible models.
• Knowledge of state-of-the-art methods coupled with the creativity and intelligence to advance beyond them.

Preferred competencies:

•   Hands on experience in SQL

• Working knowledge of AWS Sagemaker
• Masters or Doctorate (PhD) in NLP

• Publications or demonstrable strong technical results in a specialized subfield of NLP
• Strong understanding of Transformer architecture and its variants, familiarity with continue pre-training and fine-tuning language model with huggingface framework
•   Domain knowledge in insurance, financial services, advertising, or marketing industry is preferred but not required.
•   Strong understanding of Transformer architecture and its variants, familiarity with pre-training and fine-tuning language models with huggingface framework.
• Working knowledge of AWS, Spark, and data warehousing.
• Experience authoring research papers and getting those papers published.

To apply:

•   Resume
•   Link to Github

Our Pack at Embroker lives our values

• Pack First
  We succeed and fail as one team. We always optimize for what is best for our entire organization. We communicate honestly and openly, treat each other with mutual respect, and assume positive intent in interactions. 

• Create Magic
  We deliver delightful experiences at every customer touchpoint and dedicate ourselves to make each one exceptional. We build transformational world-class products by applying our full creativity to find solutions to even the hardest problems. 

• Be All-In
  We make focused commitments. We are accountable to ourselves and each other to deliver on time. We move fast and attack challenges with relentless positivity. We build things that make us proud.

We believe that systemic structures and practices disproportionately disadvantage the most marginalized people in society — including people of color, people from working-class backgrounds, women, and LGBTQ people. We believe that these communities must be represented and included in the work we do, to make our Pack stronger, more creative, and improve the way we do business. We strongly encourage applications from people with these identities or who are members of other marginalized communities.