Work · Neeraj Rondla

—/All work

Every project.

Eight shipped products. Each accent color identifies a different project — from AI coding infrastructure to health platforms to full-stack e-commerce.

Creator

CRE Intelligence Platform

NYC Agentic Underwriter & Feasibility

Built an autonomous CRE underwriting platform resolving NYC street addresses via DCP GeoSearch. Integrates parallel DoITT PLUTO & ACRIS Deed Sales queries and a Zoning RAG vector database. Features a pure CSS 3D buildable massing visualizer, interactive pro forma calculator, and a Claude-powered 'Ask the Agent' chat drawer.

ReactFastAPIPythonClaude SonnetLeaflet

Autonomous commercial real estate underwriting platform mapping NYC comparable sales and zoning limits, dropping feasibility report latency to ~4 seconds.

Case Study

Creator & Maintainer

VIBETTER

Open-Source MCP Server for AI Coding IDEs

Built and maintained an open-source MCP server that brings deep codebase intelligence to AI coding IDEs — git-diff explanations, source-grounded Q&A with file:line citations, dependency graph visualization, and audio code walkthroughs.

PythonFastMCP 3.0Gemini APIVue FlowWatchdog

Open-source MCP server powering codebase-intelligent AI workflows in Claude Code, Cursor, and Windsurf.

Case Study

Creator & Maintainer

Context Refinery

LangGraph Multi-Agent Context Orchestration Engine

Built a cross-platform desktop app (Tauri v2 + Vue 3 + FastAPI) with a LangGraph multi-agent pipeline — intent analysis, hybrid retrieval (ChromaDB dense + BM25 sparse + Reciprocal Rank Fusion), cross-encoder reranking, and iterative self-refinement — with multi-provider LLM support including local models via Ollama.

PythonLangGraphFastAPIVue 3Tauri v2

Cross-platform desktop engine that turns raw sources into clean, reranked, self-refined LLM context through a four-stage agentic pipeline.

Case Study

AI Software Engineer

Production RAG System

Enterprise Knowledge Retrieval & LLM Evaluation at T-Mobile

Built a production RAG pipeline for customer support knowledge retrieval — from document ingestion and recursive chunking to embedding, vector indexing, and semantic search with metadata filtering. Paired with an LLM evaluation framework that measurably reduced factual errors.

PythonLangChainChromaDBFastAPIRAGAS

Sub-2-second retrieval across 10,000+ pages with 60% reduction in hallucinations through systematic evaluation.

Case Study

AI Engineer

Autonomous AI Agent

Multi-Step Reasoning with Tool Use & Cited Reports

Built a multi-step ReAct agent with GPT-4 function calling that autonomously searches the web, synthesizes findings across sources, and generates cited research reports — with dual memory, self-correction loops, and a streaming frontend.

PythonLangChainGPT-4 Function CallingFastAPIVue.js

Autonomous agent that searches, synthesizes, and generates cited research reports with 5–10 tool calls per query.

Case Study

2025