Open to opportunities

venkata sai

@venkatasai

Message

I build agentic AI platforms for customer support, code review, and research workflows.

Zimbabwe

Message

I built AI platforms for D2C customer support, live code review, and research automation using Python, FastAPI, LangGraph, retrieval systems, and knowledge graphs. My AI Customer Support Agent System delivered 2–4s end-to-end latency, improved R@1 from 25.6% to 66.7%, and added safeguards for PII, prompt injection, hallucinations, and sensitive actions.

With CodeTurtle, I shipped a local AI code-review platform for GitHub repositories and live pull requests, indexing 2.5K–3.7K+ files and mapping 350+ dependency edges in Neo4j. I also developed Helix Research, an automated research pipeline that reduced collection-query latency from approximately 40 seconds to 4 seconds through retrieval, caching, model routing, and evidence consistency checks.

Experience

Work history, roles, and key accomplishments

AI Customer Support Agent System

Projects

Jun 2026 - Present (2 months)

Designed a 7-stage agent workflow for D2C support with persistent case memory, integrating Shopify and Stripe APIs for high-risk actions. Enhanced hybrid retrieval using Pinecone, BM25, and reranking, improving retrieval metrics significantly.

Python fastAPI LangGraph Pinecone Ollama Shopify Stripe RAG Hybrid Search Langsmith Prometheus Grafana

Education

Degrees, certifications, and relevant coursework

VNR Vignana Jyothi Institute of Engineering and Technology

Bachelor of Technology, Engineering

2022 -

Activities and societies: Executive Head – IE(I) Student Chapter (2024): Led technical initiatives and coordinated work for 100+ students. Organizer – TEDx VNRVJIET (2023): Managed speaker coordination, logistics, and event execution.

Pursuing a Bachelor of Technology degree, expected to graduate in 2026.