Machine Learning Engineer - GenAI, RAG & Recommendations (2+ Years)
Roles and Responsibilities :
- Build and deploy scalable LLM-based systems using OpenAI, Claude, LLaMA, or Mistral for contract understanding and legal automation.
- Design and implement Retrieval-Augmented Generation (RAG) pipelines using vector databases (FAISS, Pinecone, Weaviate).
- Fine-tune and evaluate foundation models for domain-specific tasks like clause extraction, dispute classification, and document QA.
- Create recommendation models to suggest similar legal cases, past dispute patterns, or clause templates using collaborative and content-based filtering.
- Develop inference-ready APIs and backend microservices using FastAPI / Flask, integrating them into production workflows.
- Optimize model latency, prompt engineering, caching strategies, and accuracy using A / B testing and hallucination checks.
- Work closely with Data Engineers and QA to convert ML prototypes into productionready pipelines.
- Conduct continuous error analysis, evaluation metric design (F1, BLEU, Recall@K), and prompt iterations.
- Participate in model versioning, logging, and reproducibility tracking using tools like MLflow or LangSmith.
- Stay current with research on GenAI, prompting techniques, LLM compression, and RAG design patterns.
Qualifications :
Bachelor's or Master's degree in Computer Science, AI, Data Science, or a related field.2+ years of experience in applied ML / NLP projects with real-world deployments.Experience with LLMs like GPT, Claude, Gemini, Mistral, and techniques like fine-tuning, few-shot prompting, and context window optimization.Solid knowledge of Python, PyTorch, Transformers, LangChain, and embedding models.Hands-on experience integrating vector stores and building RAG pipelines.Understanding of NLP techniques such as summarization, token classification, document ranking, and conversational QA.Bonus :
Experience with Neo4j, recommendation systems, or graph embeddings.(ref : hirist.tech)