Generative AI Engineer - Python/FastAPI

Role Overview:

Were looking for a GenAI Developer to design, build, and deploy intelligent LLM-powered systemsfrom single-agent chatbots, copilots to complex multi-agent applicationsat scale. We are particularly interested in candidates who have hands-on experience in taking GenAI applications from concept to production, especially within high-volume B2C environments. This role prioritizes individuals who understand the nuances of deploying, maintaining, and optimizing GenAI solutions for real-world users, beyond the scope of Proof-of-Concept (PoC) development. You will work across the full stack, integrating LLMs, microservices, vector databases, backend APIs, and modern cloud infrastructure.

Key Responsibilities:1. GenAI Application Development & Deployment:

• Develop scalable, asynchronous microservices using Python (FastAPI) for chatbots, copilots, and agentic workflows.

• Design event-driven architectures to support high concurrency, rate limiting, and real-time responsiveness.

• Implement secure, versioned REST/gRPC APIs

• Use Pydantic, dependency injection, and modular coding practices for maintainability.

• Proficient in working with databases using ORMs like SQLAlchemy

• Ensure observability using logging, metrics, tracing, and health checks.

• Create responsive React.js frontends integrated via REST APIs or WebSockets.

• Deploy applications on Cloud Run, GKE, using Docker, Artifact registry, CI/CD pipelines

2. LLM-Powered Conversational Interfaces:

• Design and build LLM-powered chatbots, voicebots, copilots and other applications using LangChain or custom orchestration frameworks.

• Integrate enterprise-grade LLM APIs (Gemini, OpenAI, Claude) for multi-turn, intelligent interactions.

• Implement user session management and context/state tracking for personalized and continuous conversations.

• Build RAG pipelines with vector databases, knowledge graphs to ground responses with external knowledge and documents.

• Apply advanced prompt engineering (ReAct, Chain-of-Thought with tool calling) for precise and goal-oriented outputs.

• Ensure performance in low-latency, streaming environments using WebSockets, gRPC, and SIP media gateways.

• Perform fine-tuning of open-source LLMs (LLaMA variants) using techniques like SFT, LoRA, for cost-effective domain adaptation.

• Optimize high-speed inference pipelines leveraging multi-GPU clusters (up to 8x H100s) to reduce latency and improve throughput.

3. Multi-Agent Systems & Orchestration: