Python Developer

Experience Level:

Employment Type:

• Design, architect, and implement scalable multi-tenant backend systems for RAG-based AI agents, ensuring strict data isolation across users.
• Build and optimize data ingestion pipelines for various file formats (PDF, DOCX, XLSX, TXT), including text extraction, chunking, and embedding generation.
• Develop semantic search capabilities with multilingual support, handling spelling variations, relativity, and cross-language queries (e.g., Hindi queries on English data).
• Integrate and orchestrate LLM calls using user-provided API keys (primarily OpenAI, with abstraction for future providers like Anthropic/Google).
• Create custom agent frameworks allowing users to configure tone, behavior, responsibility, message length, and model selection; ensure agents work efficiently on cost-effective models (e.g., gpt-3.5-turbo or gpt-4o-mini).
• Implement advanced RAG flows: vector retrieval (with relevance thresholds), context-aware conversation history, and fallback mechanisms (e.g., clarifying questions when knowledge base match is low).
• Manage data expiration, async processing (uploads/embedding), and scheduled cleanup jobs.
• Ensure production-grade scalability: Horizontal scaling, async queues, caching, rate limiting, and deployment on cloud (AWS EKS/Kubernetes preferred).
• Enforce security best practices: Encryption, JWT auth, no cross-user data exposure.
• Optimize prompts and chains for reliability on cheaper models; integrate internal embedding models (e.g., Hugging Face multilingual-e5-large).
• Collaborate on full system: APIs (FastAPI), databases (PostgreSQL + pgvector), storage (S3), queues (Celery/RabbitMQ/Redis).
• Mentor junior developers and contribute to architecture decisions for future extensions (e.g., other LLM providers).

Required Skills & Experience

• Core Backend:

   6+ years with Python; expert in FastAPI (or similar async frameworks like Django/Flask with migration experience).

• AI/LLM Expertise:

   2+ years building production RAG systems; hands-on with LangChain/LangGraph, vector databases (pgvector, Pinecone, Qdrant, or similar), embeddings (Hugging Face Transformers).

• Scalability & Production:

   Proven experience deploying scalable, multi-tenant applications (millions of users/data points); Kubernetes/EKS, Docker, Celery/RabbitMQ/Redis, async processing.

• Databases & Storage:

   Strong with PostgreSQL (including extensions like pgvector), SQLAlchemy; experience with S3/MinIO for file storage.

• AI-Specific:

   Deep understanding of RAG pipelines (chunking, embedding, retrieval, hybrid search); prompt engineering for cheap/efficient models; handling multilingual semantic search and context awareness.

• Security & Isolation:

   Experience enforcing tenant isolation (metadata filtering, no cross-joins).

• Tools & Frameworks:

   LiteLLM (or similar for LLM abstraction), PyPDF2/python-docx/pandas for file processing.

• Cloud:

   Hands-on with AWS (S3, EKS, ECR) or equivalent (GCP/Azure).

• Soft Skills:

   Strong problem-solving, ability to own end-to-end features, experience in production monitoring (Prometheus/Sentry).

Nice-to-Have

• Contributions to open-source AI projects or relevant GitHub portfolio.
• Experience with agentic workflows, fine-tuning, or evaluation frameworks.
• Familiarity with frontend integration (though focus is backend).
• Knowledge of cost optimization in LLM usage.