GenAI Technical Architect -Technical Lead-App Development

About Birlasoft

Key Responsibilities

• Design and Architecture: Create scalable and modular architecture for GenAI applications using frameworks like Autogen, Crew.ai, LangGraph, LlamaIndex, and LangChain.
• Python Development: Lead the development of Python-based GenAI applications, ensuring high-quality, maintainable, and efficient code.
• Data Curation Automation: Build tools and pipelines for automated data curation, preprocessing, and augmentation to support LLM training and fine-tuning.
• Cloud Integration: Design and implement solutions leveraging Azure, GCP, and AWS LLM ecosystems, ensuring seamless integration with existing cloud infrastructure.
• Fine-Tuning Expertise: Apply advanced fine-tuning techniques such as PEFT, QLoRA, and LoRA to optimize LLM performance for specific use cases.
• LLMOps Implementation: Establish and manage LLMOps pipelines for continuous integration, deployment, and monitoring of LLM-based applications.
• Responsible AI: Ensure ethical AI practices by implementing Responsible AI principles, including fairness, transparency, and accountability.
• RLHF and RAG: Implement Reinforcement Learning with Human Feedback (RLHF) and Retrieval-Augmented Generation (RAG) techniques to enhance model performance.
• Modular RAG Design: Develop and optimize Modular RAG architectures for complex GenAI applications.
• Open Source Collaboration: Leverage Hugging Face and other open-source platforms for model development, fine-tuning, and deployment.
  

Required Skills

• Python Programming: Deep expertise in Python for building GenAI applications and automation tools.
• LLM Frameworks: Proficiency in frameworks like Autogen, Crew.ai, LangGraph, LlamaIndex, and LangChain.
• Large-Scale Data Handling & Architecture: Design and implement architectures for handling large-scale structured and unstructured data.
• Multi-Modal LLM Applications: Familiarity with text chat completion, vision, and speech models.
• Fine-tune SLM(Small Language Model) for domain specific data and use cases.
• Prompt injection fallback and RCE tools such as Pyrit and HAX toolkit etc.
• Anti-hallucination and anti-gibberish tools such as Bleu etc.
• Fine-Tuning Techniques: Mastery of PEFT, QLoRA, LoRA, and other fine-tuning methods.
• LLMOps: Strong knowledge of LLMOps practices for model deployment, monitoring, and management.
• Responsible AI: Expertise in implementing ethical AI practices and ensuring compliance with regulations.
• RLHF and RAG: Advanced skills in Reinforcement Learning with Human Feedback and Retrieval-Augmented Generation.
• Modular RAG: Deep understanding of Modular RAG architectures and their implementation.
• Hugging Face: Proficiency in using Hugging Face and similar open-source platforms for model development.
• Front-End Integration: Knowledge of front-end technologies to enable seamless integration of GenAI capabilities.