Description :

Profile Summary :

We are seeking a Senior LLM Engineer with deep expertise in transformer-based NLP models (GPT, BERT, T5, RoBERTa, etc.) and a strong command of prompt engineering, fine-tuning, and instruction-based learning. The ideal candidate will design, optimize, and deploy large language models (LLMs) for real-world applications in text generation, summarization, classification, and reasoning. This role requires a balance of research depth and engineering excellence, with hands-on experience in Python, Hugging Face Transformers, and deep learning frameworks such as PyTorch or TensorFlow.

Roles and Responsibilities :

- Develop, fine-tune, and optimize transformer models (GPT, BERT, T5, RoBERTa, etc.) for multiple NLP tasks (summarization, classification, translation, question answering).

- Perform domain-specific fine-tuning of pre-trained models to enhance contextual accuracy and performance.

- Design and iterate prompts and instruction-based strategies for guiding LLM behavior across use cases.

- Apply zero-shot, few-shot, and many-shot learning to improve model adaptability and performance.

- Implement Chain-of-Thought (CoT) prompting for improved reasoning and structured output.

- Evaluate models using BLEU, ROUGE, perplexity, and other key performance metrics.

- Deploy fine-tuned and optimized models into production pipelines with scalability and monitoring in mind.

- Identify, analyze, and mitigate biases, hallucinations, and factual inaccuracies in LLM outputs.

- Collaborate closely with data scientists, ML engineers, and product teams to deliver AI-driven solutions.

- Contribute to continuous improvement by staying updated with advances in LLM architectures, prompting methods, and evaluation techniques.

Mandatory Skills & Technical Proficiency :

- Transformer-based models (GPT, BERT, T5, RoBERTa, etc.), attention mechanisms, embeddings, tokenization, context windows

- Fine-tuning, prompt engineering, zero/few/many-shot learning, instruction-based prompting, Chain-of-Thought reasoning

- Python, Hugging Face Transformers, PyTorch, TensorFlow, SpaCy, NLTK

- BLEU, ROUGE, perplexity, accuracy, F1-score

- Model deployment into production (APIs, real-time pipelines), monitoring for drift and reliability

- Bias detection, hallucination control, model interpretability

- Git, JIRA, CI/CD workflows for ML pipelines

Good to Have :

- Experience with LLM observability and monitoring tools.

- Exposure to cloud platforms (AWS, GCP, Azure) for scalable deployments.

- Understanding of MLOps principles for full model lifecycle management.

- Familiarity with vector databases, embedding stores, and retrieval-augmented generation (RAG).

Education :

- Bachelors or Masters degree in Computer Science, Artificial Intelligence, Data Science, or related field, or equivalent practical experience.