Generative AI Developer / Architect - LLM

Description : Role Summary :

We are seeking an experienced Generative AI Developer / Architect with a strong track record of building and deploying LLM-powered applications in real-world environments. The ideal candidate will bring deep technical expertise in LLMs, prompt engineering, RAG architectures, and cloud-based AI services (Azure OpenAI, AWS Bedrock).

This is a high-impact role for someone who thrives at the intersection of innovation, scale, and responsible AI. Youll be driving the end-to-end design, development, and deployment of production-grade GenAI solutions for diverse enterprise use cases.

Key Responsibilities :

• Design and develop GenAI applications using state-of-the-art LLMs such as GPT (OpenAI / Azure), Claude (Anthropic), LLaMA (Meta), etc.
• Build prompt engineering pipelines, including few-shot, chain-of-thought, and role-based prompts for enhanced context understanding.
• Develop RAG (Retrieval-Augmented Generation) pipelines using vector databases for dynamic knowledge retrieval from enterprise data.
• Architect and implement agent-based GenAI workflows for multi-step tasks, automation, and decision-making systems.
• Deploy scalable GenAI applications on Azure OpenAI (AI Studio) or AWS Bedrock, leveraging managed services and serverless infrastructure.
• Utilize Python, FastAPI, and frameworks like LangChain, LLamaIndex, or Haystack for backend orchestration.
• Integrate vector databases such as FAISS, Pinecone, Weaviate, or Chroma for efficient embedding storage and similarity search.
• Implement cost optimization strategies by identifying high-impact GenAI use cases and minimizing token consumption.
• Enforce Responsible AI practices, including fairness, explainability, and privacy compliance across all GenAI deployments.
• Design controls for prompt injection prevention, jailbreaking mitigation, and output filtering using input / output sanitization.
• Build enterprise Q&A systems using embedded knowledge bases, PDFs, internal wikis, and structured databases.
• Implement Human-in-the-Loop (HITL) mechanisms for validation, continuous learning, and human oversight.
• Design multimodal pipelines (text + image + voice) and handle real-time parsing, transcription, chunking, and token management.

Required Skills & Experience :

(ref : hirist.tech)