Direct message the job poster from Zodiac Solutions, Inc
Krishna K.
Sr. Account Manager- Client Engagement Management
Job Description:
Develop and fine tune NLP/NLU models, including transformer based models (BERT, GPT etc.,) for a range of language understanding tasks (language generation, sentiment analysis, entity recognition etc.,)
Additionally design and implement generative AI models by leveraging LLMs for text generation and understanding.
Preprocess and curate text data, including text cleaning, tokenization and data augmentation to prepare it for training NLP/NLU models
Innovate and implement cutting edge algorithms to improve the performance, accuracy and efficiency of the NLP/NLU models
Assess model performance using relevant NLP metrics and optimize models for real world application
Stay updated with latest developments in Generative AI specifically on NLP/NLU research and apply new techniques to enhance our solutions
Strong expertise in NLP/NLU with focus on deep learning models and open source LLMs such as Llama 2, Falcon etc.,
Proficiency in programming languages such as Python and experience with NLP libraries/frameworks like spaCy, NLTK, Hugging Face Transformers and Generative AI and LLM frameworks
Familiarity with cloud computing platforms for model training and deployment
Experience with tools like Langchain, Embedchain and Vector Database (Pinecone, FAISS, Milvus, ChromaDB, Vespa etc.,) to enhance data processing and model development workflow as well as rich understanding of vector stores and search algorithms
Exceptional problem solving skills and the ability to work on complex language understanding and Generative AI projects
Experience designing control and sandboxing systems for AI research
Hands on experience with Pytorch, TensorFlow
Direct engineering experience of high performance, large-scale ML systems
Hands on MLOps experience, with an appreciation of the end-to-end CI/CD process
Have experience supporting fast-paced startup engineering teams
Experience maintaining and/or contributing to bug bounty and responsible disclosure programs