NLP Algorithm Development on AWS SageMaker

Job Title: AI/ML Developer for Ingredient Comparison & Clean Product Scoring Algorithm About Us: We’re seeking an experienced AI/ML developer or development team to create an algorithm that can evaluate and compare ingredients in various products, identifying the ones with the cleanest ingredients based on specific criteria. This project will involve developing a Natural Language Processing (NLP)-based model to assess ingredient lists against a database of defined cleanliness standards. The final output should allow end users to evaluate product cleanliness and make informed choices. Project Overview: The goal of this project is to build a platform that can: 1. Ingest ingredient lists from consumer products. 2. Use AI/ML and NLP to evaluate each ingredient against predefined criteria for “cleanliness.” 3. Output a cleanliness score and summary, highlighting potentially harmful or undesirable ingredients. Responsibilities: • Design and develop an NLP-based algorithm to evaluate ingredient lists. • Build or integrate a database of ingredients, labeling each by safety and cleanliness standards. • Implement a scoring system based on the “clean” criteria defined. • Create a dashboard or user interface that displays results, including the cleanliness score and detailed ingredient analysis. • Ensure scalability and ease of adding new ingredients or updating cleanliness criteria. Key Deliverables: 1. Cleanliness Scoring Algorithm: An NLP-based model to classify ingredients into categories based on predefined cleanliness standards. 2. Ingredient Database: A structured database containing ingredient names, types, sources, and cleanliness scores, with the flexibility for updates. 3. User Interface (UI): An intuitive UI displaying results for users, including the cleanliness score and detailed analysis. 4. Documentation and Testing: Full documentation of the model, database, and UI, along with testing to ensure accuracy and functionality. Technical Requirements: 1. Machine Learning & NLP: • Develop a custom NLP model (or fine-tune an existing transformer model like BERT) to process and classify ingredients based on defined cleanliness criteria. • Implement Named Entity Recognition (NER) to identify and normalize ingredient names. • Use language processing to handle variations in ingredient naming and match them accurately to the database. 2. Database Structure: • Build or integrate an ingredient database with key attributes: • Ingredient Name • Category (e.g., preservative, fragrance) • Cleanliness Score (out of 10 based on criteria like toxicity, environmental impact) • Source (synthetic, natural, animal-derived) • Known Issues (e.g., carcinogenic, allergen) • Make the database easily updatable, allowing new ingredients or criteria adjustments as needed. 3. Scoring System: • Develop a scoring algorithm to evaluate product cleanliness by aggregating individual ingredient scores. • Allow for customized weighting based on factors like toxicity, environmental impact, and animal testing. • Provide flexibility for users to set personal preferences (e.g., avoid all animal-derived ingredients). 4. Front-End Interface (Optional): • Build a simple, intuitive dashboard for users to input product information and view cleanliness results. • Display color-coded ingredient lists and an overall cleanliness score. • Optionally, include a “flag” feature to highlight harmful ingredients. 5. Tech Stack: • Preferred languages: Python for machine learning and NLP tasks; JavaScript or React for the front end. • Libraries: spaCy or Hugging Face Transformers for NLP, Pandas for data processing, and Scikit-learn or TensorFlow for machine learning. • Database: SQL or NoSQL database, with potential integration with open-source databases like EWG or COSING. • Hosting: AWS, Google Cloud, or other cloud service providers for scalable deployment. Qualifications: • Proven experience with machine learning, NLP, and database design. • Familiarity with ingredient databases, cosmetic or consumer product standards, and cleanliness criteria is a plus. • Proficient in using frameworks like TensorFlow, PyTorch, spaCy, or Hugging Face Transformers. • Experience building user interfaces and visualizing AI output in a way that’s accessible for end users. • Strong communication skills and experience with full-cycle development. To Apply: Please submit your application along with: 1. A brief proposal outlining your approach to this project. 2. Examples of similar NLP or AI projects you’ve completed. 3. Estimated timeline and budget for this project.

Project ID: 38747004