Junior Retrieval-Augmented Generation And Large Language Models Research Engineer

What role will you play in the team

You will play a foundational role in our core Data Science, Analytics & Machine Learning department by designing, fine-tuning, and deploying advanced Large Language Models and Retrieval-Augmented Generation architectures. Your work will directly empower our enterprise platforms to ingest massive corporate knowledge bases, execute precise semantic searches, and deliver context-aware generative outputs with exceptionally low latency and high factual accuracy across production environments.

What you will do

You will architect end-to-end vector database pipelines, integrate state-of-the-art embedding models, and optimize proprietary prompt engineering strategies using cutting-edge frameworks. Collaborating closely with senior data scientists and machine learning engineers, you will establish robust evaluation benchmarks, implement stringent hallucination mitigation protocols, and deploy scalable inference endpoints leveraging OpenAI API, Claude API, and open-source foundation models.

Key responsibility

• Design, implement, and optimize Retrieval-Augmented Generation (RAG) pipelines utilizing advanced vector databases like Pinecone, Milvus, or ChromaDB for efficient document retrieval.
• Fine-tune and deploy Large Language Models (LLMs) using PyTorch, Hugging Face Transformers, and parameter-efficient fine-tuning techniques such as LoRA and QLoRA.
• Develop automated data ingestion and chunking strategies to parse unstructured enterprise documents, PDFs, and relational databases for optimal embedding generation.
• Conduct rigorous prompt engineering iterations, implementing few-shot learning paradigms and chain-of-thought reasoning to enhance generative output reliability.
• Collaborate with MLOps engineers to containerize models using Docker and orchestrate scalable deployment pipelines on cloud infrastructure with GPU acceleration.
• Monitor model drift, latency, token consumption, and generation accuracy in production environments, applying continuous optimization methodologies.
• Build custom evaluation frameworks leveraging RAGAS and TruLens to quantitatively measure context precision, faithfulness, and answer relevance.

Required Qualification and Skills

• Bachelor's or Master's degree in Computer Science, Artificial Intelligence, Data Science, or a highly quantitative discipline from a recognized institution.
• Demonstrated hands-on experience of 0 to 1 years in building and deploying machine learning models, with explicit focus on NLP, LLMs, and RAG architectures.
• Proficiency in Python programming, along with strong command over core data manipulation libraries such as Pandas, NumPy, and LangChain.
• Solid understanding of deep learning fundamentals, neural network architectures, attention mechanisms, and transformer-based models using PyTorch.
• Hands-on familiarity with vector search technologies, embedding generation models, and API integrations including OpenAI API, Claude API, and Hugging Face.
• Familiarity with database management systems including PostgreSQL, MySQL, and NoSQL databases like MongoDB for metadata storage.
• Strong analytical mindset, exceptional debugging capabilities, and the ability to articulate complex algorithmic concepts clearly to cross-functional teams.

A Day in the Life

Your day begins by reviewing the performance telemetry of our production RAG pipelines deployed across cloud clusters. You spend the morning optimizing embedding chunk sizes in Python and experimenting with novel prompt engineering templates to reduce hallucination rates in customer support bots. After a collaborative sync with the MLOps team, you dedicate your afternoon to fine-tuning a custom transformer model using PyTorch and running automated evaluation suites using RAGAS to ensure compliance with enterprise accuracy standards before wrapping up your tasks.