Srikanth B S

AI Automation Engineer • Python Backend Developer • Full Stack Engineer • Solution Architect

I build LLM-powered automation platforms — intent analysis engines that turn support tickets into reusable playbooks, and self-updating knowledge wikis with RAG assistants. Durable pipelines with Temporal, semantic search on PostgreSQL + pgvector, and React front-ends that make it all usable.

0 Years Experience

0 Major Projects

0 % Uptime

⚡ Flagship · LLM Automation

Intent Analysis Automation

Turns raw e-commerce customer-support conversations — every incoming and outgoing ticket message — into a reusable service playbook. Each message is classified for intent and response type, and the analysis layer generates the triggers, response templates, and decision trees that tell any customer-service team exactly which process to follow.

Durable pipeline on Temporal — ingest → normalize → embed → classify → aggregate, with automatic retries and fault-tolerance
Embeddings + Gemini 3.1 Pro classify every message's intent and required response type
Auto-generated playbook — triggers (when to act), templates (what to send), decision trees (which path per intent × response type), versioned in PostgreSQL
React analytics dashboard — intent explorer, template library, decision-tree viewer for support analysts
Reusable reference — any other customer-service operation can adopt the proven playbook

Python PostgreSQL Temporal Embeddings Gemini 3.1 Pro ReactJS

Intent Analysis Automation — system architecture: helpdesk tickets flow through a Temporal pipeline (normalize, embed, Gemini classification, aggregate) into a PostgreSQL service playbook of triggers, templates and decision trees, surfaced in a React dashboard

View the decision workflow

Workflow: a ticket message is classified for intent and response type, routed through the decision tree, and answered via trigger + template

📚 Flagship · RAG Platform

Knowledge Wiki + Ask Wiki

Connects a company's document drive via OAuth, watches the folder they grant, and turns every document into a Wikipedia-style page — side menus, sections, cross-links — kept in sync on every addition, update, or deletion. Ask Wiki lets the company's support staff query that knowledge in real time and get grounded answers with citations.

Company OAuth logins — folder-scoped drive access granted by each customer, fully multi-tenant
Change tracker — additions create pages, edits re-sync them, deletions retire them: zero-touch
Temporal doc pipelines — parse → chunk → embed → generate pages → cross-link & index
PostgreSQL + pgvector stores pages, chunks and embeddings for semantic retrieval
Ask Wiki (RAG) — on-time LLM answers grounded strictly in the company's wiki docs, with source citations