AI Engineer (RAG/NLP) Gurgaon (On-site)
About The Role
We are seeking an experienced AI Engineer to design, build, and deploy advanced Generative AI and NLP solutions, with a focus on Retrieval-Augmented Generation (RAG) pipelines, document automation (OCR/ASR), and knowledge-assist systems. The ideal candidate will have strong hands-on experience with Python, Transformers, vector databases, and API deployment, and will be comfortable managing the entire lifecycle of AI models, from data preparation to monitoring in production.
Key Responsibilities
- Design, build, and optimize RAG pipelines, including prompting, text chunking, retrieval, reranking, and evaluation using vector databases such as Chroma DB or Qdrant, and Transformer-based LLMs like Llama, Mistral, or BERT family models.
- Productionize ASR systems (e.g., Whisper-large-v3) for call centre and voice-based use cases, ensuring improvements in both accuracy and latency.
- Develop OCR and document digitization workflows using tools such as OpenCV and Tesseract, along with CNN/LSTM-based post-processing for unstructured PDFs and images.
- Build and deploy APIs using Fast API or Flask, integrating with existing services and data sources such as MongoDB.
- Orchestrate data and model workflows using Airflow, automating ETL processes, evaluation pipelines, and periodic retraining.
- Implement CI/CD pipelines for model and API releases, ensuring strong testing, logging, and observability practices.
- Manage both offline and online evaluation for metrics such as latency, accuracy, F1 score, ASR WER, and retrieval precision/recall, and provide detailed analytical reports and recommendations.
- Collaborate closely with product and operations teams to translate business challenges into measurable ML objectives and service-level agreements (SLAs).
Must-Have Skills
- Proficiency in Python (production-grade), PyTorch, and Hugging Face Transformers.
- Strong understanding of RAG fundamentals, including text chunking, embedding selection, retrieval (dense and sparse), reranking, and evaluation frameworks.
- Experience with vector search and data stores such as Chroma DB or similar technologies, along with solid data modelling and indexing expertise.
- Practical knowledge of API development using Fast API or Flask, including RESTful best practices, authentication, rate limiting, and pagination.
- Experience in MLOps using Docker, CI/CD, Linux, Git, and tools for logging and monitoring model services.
- Exposure to OCR and ASR systems, particularly OpenCV, Tesseract, and Whisper (or equivalent frameworks).
- Strong grasp of classical NLP and ML techniques, including tokenization, LSTMs/CNNs, XGBoost, and metric-driven Skills :
- Experience fine-tuning large language models or encoders for classification, summarization, and domain adaptation.
- Understanding of prompt engineering, tool integration, and evaluation for LLM-based applications.
- Familiarity with scaling retrieval systems for low-latency, high-availability production environments.
- Experience with document question answering, email or call centre analytics, or enterprise knowledge management.
- Knowledge of agentic AI frameworks such as Lang Chain, Lang Graph, or Crew AI.
Qualifications
- 2 to 5 years of hands-on experience in AI, ML, or NLP engineering with proven production ownership.
- Bachelors degree in computer science, or a related field, or equivalent practical experience.
- Strong portfolio or GitHub profile demonstrating shipped APIs, model implementations, and well-documented repositories with test coverage.
(ref:hirist.tech)
