WorkSkillsRecognitionAboutContactRésumé ↓
AI Product Management

I build AI products,then prove they work.

An engineer moving into product management, focused on AI-native products and the platform layer of B2B SaaS. Four years building the billing and self-serve infrastructure beneath Freshworks' AI products. Now I ship AI systems, stress-test my own claims against ground truth, and publish what breaks.

Open to product roles across B2B SaaS, FinTech, B2C, and AI-native teams.

Open to opportunities·Bengaluru, IndiaEmailLinkedIn
The AI-native loop

As AI makes building a commodity, the job moves upstream, to judgment and proof.

Discover
user signals, synthesized
Prioritize
trade-offs, the right bet
Align
PRD, context, direction
Ship
agentic tools, prototypes
Prove
evals vs ground truth
learning feeds the next cycle
evidence
§ 01

Selected work

01
GroundTruth
Agentic AI System · Multi-Agent Orchestration
A grounded multi-agent support system that answers only when every claim traces to a retrieved source, and refuses when it cannot.
Verification caught a false 90% match
02
Receipts
AI Tool · Product
Turns raw interview transcripts into synthesis where every theme ties back to the exact quote that supports it.
100% fidelity, 0% hallucination (n=6)
03
FitSense
AI Product · Prototype
An AI fit-advisor that answers “will this actually fit me?” at the moment of purchase, not after a return.
Solved at the point of purchase
04
Freddy AI, A Strategic Teardown
AI Teardown
A PM's read of Freshworks' three strategic bets at Refresh 2026, and the data that would validate each one.
Three bets, examined honestly
05
Where the bottleneck moved
Research Synthesis
Four senior engineers spanning Razorpay, WEX, Uber, Freshworks, and Target on where the slow step went once AI took over the coding.
Four engineers, four answers
§ 02

Skills & tools

AI / ML for Products
LLMs RAG Agentic AI AI-agent design Harness engineering Loop engineering Prompt & tool design Evals & groundingLangChainLangGraph Lyzr AIn8n Emergent OpenAI APIClaudeClaude Code AWS Bedrock GPT-4oGemini Lovable GammaMLflowHugging FaceWeights & Biases
Product Management
PRDs Working-backwards docs Requirement elicitation User stories RICE MoSCoW Agile / Scrum AARRR Competitive teardown GTM & pricing
Analytics & Experimentation
Mixpanel A/B testing Cohort analysis Event instrumentation KPI dashboards
Technical Foundation
ReactJavaScriptTypeScript REST APIsJava / SpringStrapi v5SupabaseJest / RTLGit Azure SSO Chargebee
Design & Collaboration
Figma Aha!JiraPostmanNotionMiro
§ 03

Workshop & recognition

IISc Bangalore · Vital Skills × Rhapsody · Oct 2025

Idea to working prototype in an hour, then pitched to a room of 200+.

At a generative-AI workshop at the Indian Institute of Science, I built StyleSpot, a working AI shopping prototype, from scratch using generative-AI tools. Ideation to a running app in about an hour. Then I took the stage and pitched it live, running a full working demo of the product in front of an audience of more than two hundred experienced tech professionals and founders.

~1 hr
Idea to working
prototype
0
Tech professionals
& founders
Live
Full product demo,
on stage
StyleSpot · AI shopping assistant · built & demoed live at IISc Bangalore
LIVE PITCH · IISc BANGALOREAbhishek Katkar pitching StyleSpot live on stage at IISc Bangalore
Granted
Patent

Retro-fitment kit converting IC-engine auto-rickshaws to electric drive

A gearbox-based retrofit system to convert internal-combustion three-wheelers into electric vehicles: a first-principles hardware innovation, filed and granted.

IN 202121024842Granted Jun 2021
Award

Exceptional Individual Performance

Recognized by Freshworks for individual performance on the Platforms team, the billing and self-serve infrastructure beneath its AI products.

FreshworksPlatforms Team
proof
§ 04

About

NowAI Product Management
B2B SaaS · FinTech · B2C · AI-nativeFocusBilling · Subscriptions
Platform · AI-native surfacesPreviouslyFreshworks, Platforms team
4 yearsBasedBengaluru, India
Open to Pune · remote

For four years I built the monetization and platform layer beneath Freshworks' AI products: the billing, subscription, and self-serve infrastructure where a PM's decisions on pricing, packaging, and configuration become live product across more than ten product teams.

Software Engineer, Platforms Team · Freshworks2022 to 2026
  • $25M ARRBuilt Auto Recharge configuration tooling for Freddy AI Agent, Freshworks' usage-based AI offering; designed role-based UX across CX, EX, and Super Admin tiers, and removed CORS failures with a server-side proxy pattern.
  • 3-5 wks to minBuilt configuration and workflow tooling (Strapi v5 plugins for Auto Recharge and Addon Metadata) that let 10 to 15 PMs push live UBX changes in minutes instead of a 3-5 week engineering and multi-region deployment cycle.
  • Studio 2025Shipped buyer-side surfaces (plan comparison, checkout) for the 2025 Freddy AI Agent Studio add-on; partnered with the PM to align packaging across recurring and non-recurring billing for a usage-based AI product.

Watching that workflow collapse taught me something larger: as AI removes the engineering middle layer between PM intent and shipped product, the bottleneck moves upstream, to whoever has the judgment to choose the right thing to build. That is where I want to operate.

I made the move deliberate. Alongside applied-AI product certifications, I built and shipped the AI products in the case studies above, and stress-tested each one against ground truth.

2026Generative AI for Business Decision-Making, IIM Mumbai × MIT (Prof. David Simchi-Levi)
2026Product Management with Applied AI, IIT Roorkee (iHUB DivyaSampark)
2025Mastering Generative AI · AI with Machine Learning, IISc Bangalore
Education
B.Tech, 2017-2021 · Masai School, 2021-2022 · Languages: English, Hindi, Marathi
§ 05 · Contact

Building something in AI, billing, platform, or a product where judgment is the job? Let's talk.

Based in Bengaluru, open to relocating to Pune, and to Bangalore and remote roles. Open to product roles across B2B SaaS, FinTech, B2C, and AI-native teams.

Abhishek Katkar