Senior Data Scientist - LLM Models

Role Overview

Seeking an experienced Data Scientist with strong hands-on expertise in Generative AI, Retrieval-Augmented Generation (RAG), and Deep Learning. The role involves building, deploying, and optimizing AI/ML models to solve complex business problems, enhance automation, and develop scalable GenAI applications.

Key Responsibilities :

- Develop, fine-tune, and deploy Generative AI models (LLMs, diffusion models, transformers).

- Design and implement RAG pipelines using vector databases, embeddings, and retrieval frameworks.

- Build, train, and optimize deep learning models for NLP, computer vision, and multimodal tasks.

- Create scalable end-to-end ML pipelines for production environments.

- Conduct data preprocessing, feature engineering, and experimentation.

- Evaluate model performance with appropriate metrics and optimize for accuracy, latency, and efficiency.

- Collaborate with engineering, product, and business teams to integrate AI solutions into applications.

- Research emerging GenAI and DL techniques to enhance existing systems.

- Document architecture, workflows, and best practices.

Required Skills & Qualifications :

- Strong proficiency with Python and ML frameworks (PyTorch, TensorFlow, Keras).

- Solid understanding of LLMs, embeddings, transformers, and prompt engineering.

- Experience with RAG frameworks (LangChain, LlamaIndex, Haystack) and vector databases (FAISS, Pinecone, Milvus, Chroma).

- Hands-on experience with deep learning architectures (CNNs, RNNs, attention models).

- Familiarity with cloud platforms (AWS, Azure, GCP) and containerization (Docker, Kubernetes).

- Strong knowledge of data handling, NLP, and model deployment.

- Bachelors or Masters degree in Computer Science, AI/ML, Data Science, or equivalent.

Preferred Skills :

- Experience with MLOps tools (MLflow, DVC, Weights & Biases).

- Exposure to multimodal AI (text, image, audio).

- Experience working with LLM fine-tuning (LoRA, QLoRA, PEFT).

- Knowledge of retrieval optimizations (hybrid search, rerankers, BM25).