Hi, I'm Viraj.

I'm a product manager who builds. I've taken three products from a messy problem to launch, writing the specs, designing the screens and often shipping the code myself.

See my work

Currently Product Manager at AIRTH, open to product roles in AI, SaaS and marketplaces.

CAS demo with sample data

Neha K.

WhatsApp lead, stage: Intro
CustomerWill the split AC purifier work in a 12×14 ft bedroom? Also, where is my last order?
AI agentYes. It covers rooms up to 200 sq ft, so a 168 sq ft bedroom is well within range.
Answered from knowledge base: product specs
AI agentYour order #4821 is out for delivery and should reach you today.
Tool used: live order status
AI is handling this chat.
Click through a mock of CAS, the CRM I designed and built solo. Names and numbers here are sample data.

Selected work

Each product started with a real business problem and a tight set of limits. Here is what I built inside them.

CAS: an AI CRM built in-house

AIRTH · Designed and built end to end, solo

In daily use
Problem
AIRTH paid ₹3 lakh a year for Respond.io, yet leads, quotes, payments and orders still lived in separate tools. Customers sent 40–50 WhatsApp chats a day that the team struggled to answer quickly and correctly.
Constraints
No tech team and no budget for more software. Any AI answering customers had to reply only from verified company information, never guess a price.
What I built
  • A WhatsApp AI agent using RAG and tool calling: it reads from a knowledge base managers can edit, checks live orders and prices, and hands over to a person when unsure.
  • An auto-triggered post-sale journey: a feedback check-in after 1 month, then upsell and cross-sell after 4 months.
  • Campaign scheduling, WhatsApp template creation and automatic lead follow-ups, so all marketing runs from one place.
  • A GST-ready quotation builder that works from the dashboard or right inside WhatsApp.
Result
₹3L/yrtool cost removed
1,200+chats handled a month
8people use it daily
ReactNode.jsPostgreSQLAWS EC2 & S3OpenAIWhatsApp Cloud APIShopifyShiprocket

OPTRA ONE: field-service app

AIRTH · FDD, PRD and UI/UX, shipped with 1 SDE intern

Play Store & App Store
Problem
AC, RO and electrical service businesses assign technician jobs over phone calls. Owners lose hours deciding who goes where, and payments and renewals slip.
Constraints
Every assignment has to respect 10+ rules at once: the right skill, no double-booking, working-hour limits, travel time between jobs, leave days, urgent jobs first and the customer's preferred time. And owners won't fill long forms.
My role
  • Designed the complete UI/UX: every screen and flow for business owners and their technicians.
  • Led product engineering: wrote the FDD and PRD, defined how each feature should work including edge cases, and guided the SDE intern from spec to launch.
  • Proposed the scheduling approach: use Google OR-Tools and treat job assignment as a vehicle routing problem with time windows. I also defined the 10+ rules and the goals the scheduler balances.
Result
Live on both app stores and piloting with AIRTH's service team and channel partners to run daily field jobs.
UI/UX designPRD & FDDGoogle OR-ToolsVRP with time windowsMobile app

Warehouse management system

AIRTH · FDD, PRD and UI/UX, built with 1 intern

In daily use
Problem
Stock lived in an Excel sheet anyone could edit. Wrong numbers couldn't be traced to a person, the data got messy, and raw materials were ordered on gut feel.
Constraints
One intern developer, an operations team that needed it to be quicker than Excel, and demand that swings between summer and winter products.
What I built
  • One signed-in daily ledger per warehouse covering production, sales, returns and damage, samples, transfers, inward stock and wastage.
  • XGBoost forecasting of raw-material needs from past usage and season, so stock is ordered before it runs out.
  • Supplier quality tracking that links every raw-material receipt to its QC certificate and batch, and flags missing or failed checks.
Result
50+items tracked daily
4–5person ops team on it
XGBoostData modellingQC traceabilityDashboards

Experience

From operations and growth roles into product, always close to the data and the customer.

  1. Product Manager

    Dec 2025 – now

    AIRTH Research

    Sole tech and product owner in a non-tech company. Built CAS, OPTRA ONE and the warehouse system from problem to launch.

  2. Founder's Office, Strategy & Growth

    Jun 2025 – Oct 2025

    Harvested Robotics, agri-tech backed by Arali Ventures

    Cut manual work by 45% with GenAI workflow automations, built a YOLO weed-detection model, turned 120+ farmer interactions into insights that sped up product cycles by 25%, and saved 10–12% on procurement.

  3. Business Operations Analyst

    Dec 2024 – Jun 2025

    Sukoon Unlimited

    Raised NPS by 22% and CSAT by 18% by building and training a team of 15+ caregivers, used sentiment analysis to spot churn risk early, and cut turnaround time by 25%.

  4. Business Analyst & Marketing

    Jan 2024 – Aug 2024

    Panini AI Services

    Improved efficiency by 25% with data-driven solutions, cut process delays by 30% with new SOPs, and built a pipeline of 250+ qualified prospects.

  5. Co-Founder & President, AI Club

    Aug 2023 – May 2024

    Presidency University, Bengaluru

    Raised ₹5 lakh+ to build SmartAqua, an AI and IoT irrigation system that cut water use by 35% in field pilots, and led 10+ AI projects with a 15-person team.

Skills

Product judgement first, with enough engineering to build what I spec.

Product

  • Problem discovery
  • PRD & FDD
  • UI/UX wireframing
  • Roadmapping
  • SOP design
  • 0→1 launches

AI & ML

  • LLM agents
  • RAG
  • Tool calling
  • OpenAI API
  • AWS Bedrock
  • XGBoost
  • YOLO
  • NLP sentiment
  • Google OR-Tools

Engineering

  • React
  • Node.js / Express
  • PostgreSQL
  • DynamoDB
  • REST APIs
  • Webhooks
  • AI coding agents

AWS

  • EC2
  • Lambda
  • SQS
  • EventBridge
  • CloudWatch
  • S3

Integrations

  • WhatsApp Cloud API
  • WhatsApp Flows
  • Shopify
  • Shiprocket

Analytics & tools

  • Power BI
  • Advanced Excel
  • Zoho CRM
  • HubSpot
  • Sales Navigator

Let's talk about your product.

I'm open to product manager roles where I can own a problem end to end. The fastest way to reach me is email.

erviraj1745@gmail.comEmail linkedin.com/in/virajmagdumLinkedIn +91 93228 86397Phone Viraj_Magdum_Resume.pdfResume