Pavan A S

I’m a Computer Science undergraduate at MSRIT Bangalore, focused on building production-grade AI systems, scalable distributed backends, and full-stack applications. I’m especially passionate about agenticAI and RAG pipelines—turning complex ideas into reliable, grounded engineering outcomes.

In my AI_Invest_Agent project, I engineered a multi-step LangChain agent powered by GPT-4 to synthesize goal-aligned portfolio strategies using live market data APIs and semantic search over 10K+ SEC filings indexed in Pinecone. I designed and implemented a robust RAG pipeline with chunking, embeddings, and indexing to keep LLM responses grounded in verifiable, real-time data and reduce hallucinations. I also deployed a production-ready MERN app with a React dashboard, JWT-secured Express API, and a FastAPI AI backend, supported by cron-triggered ingestion for a continuously updated knowledge base.

I extended these ideas further with AcadenceAI, where I built a hybrid + agentic RAG academic intelligence platform using PostgreSQL + pgvector, LangChain, and GCP. I designed an adaptive, LLM-driven query system for natural language academic questions, including dynamic SQL generation, hybrid retrieval, contextual reranking, multi-step reasoning, analytics summarization, and hallucination prevention. Alongside that, I architected a scalable multi-tenant AI SaaS with JWT authentication and tenant-aware retrieval isolation, plus automation agents (including Gmail automation) for institution-scale workflows.