Resume variant

AI Automation / Agentic Systems

Tailored for AI Automation, Agentic Systems, and AI-native product roles.

AI AutomationAgentic SystemsAI-native

Summary

I build AI systems that take actions, not just answer questions. From WhatsApp → AI Operations to Fleet OS approval workflows, I design human-in-the-loop governance for production AI. My systems combine LLM integration, data harnesses, permission engines, and approval layers into working automation.

Flagship: AI Operations in Fleet OS

WhatsApp → AI Operations

WhatsApp API gateway single transport, wa.example.com gateway, human-in-the-loop approval for HIGH risk, auto-reply via [redacted phone].

Data Harness

SOURCE → IDENTITY → SCOPE → CLASSIFICATION → VALIDATION → PERMISSION → AI POLICY → ACTION → AUDIT → LEDGER. Every message gated.

OCR Processing

Cloudflare Workers + Queues + R2 + OCR. Idempotent document processing job hash(messageId+attachmentId), Evidence 96% confidence.

Group Intelligence

WhatsApp Groups → WhatsApp API → AI Classification → Operational Signals → Manager Brief → Action. LLM inference.

Other AI Systems

AI Policy & Permission Engine

User → Role → Data Scope → AI Scope → Allowed Actions. Enterprise AI governance layer.

AI Action & Approval Layer

AI assistant vs AI operator: identify → propose → approval → execute. Human-in-the-loop governance.

AI Research Worker

Sources → Research → Scoring → Synthesis → Delivery → Content. Automated research pipeline.

KYC Hybrid

local validation (Indian ID validation) + mock KYC adapter → format validated. Pluggable adapters for production KYC.

Skills

AI

AI-native workflows, Agentic systems, LLM API integration, RAG / retrieval, AI memory, AI permissions, Human-in-the-loop, Data Pipeline Design, Prompt Engineering

Engineering

Next.js, TypeScript, Cloudflare Workers, Queues, R2, Containers, WhatsApp API, Supabase, PostgreSQL

Automation

Workflow Automation, Webhooks, Event-driven Architecture, Idempotent Processing, Queue-based Workflows, Orchestration

Infrastructure

Cloudflare ecosystem, Docker, primary server, DNS, reverse proxy, firewall, warm DR, presigned URLs

Architecture

WhatsApp / Web
      ↓
Cloudflare Workers
      ↓
AI / Intelligence Layer
      ↓
Supabase / D1
      ↓
Actions + Audit

Technical decisions

Why Cloudflare Workers?

Edge execution, lightweight APIs and integration with the existing data layer. Queues for async processing.

Why LLM inference?

Low-latency inference for real-time WhatsApp classification. Sub-second response times.

Why WhatsApp API?

WhatsApp is how riders communicate. WhatsApp API gives programmatic access. Single transport.

Proof

Architecture →GitHub →Live Demo →Case Study →

Experience

AI Automation Lead — Vridhi Labs

2023 – Present • Hyderabad

Built WhatsApp → AI Operations pipeline, Data Harness, AI Policy Engine, OCR processing, Group Intelligence. Designed human-in-the-loop approval workflows for production AI systems.

Marketing Associate — MediaMint

2021 – 2023 • Hyderabad

Built automated lead routing, nurture sequences, campaign operations. Treated marketing as engineering problem.

Proof

AI systems built

96%

OCR confidence

human-in-the-loop approval

for HIGH risk actions