I'm a Product Leader who builds things. Not just roadmaps and PRDs — actual platforms, with real AI wired in, patents filed, and revenue generated.
9+ years in enterprise tech — currently Group Manager (Associate Director) & AI R&D Lead at HCL Software ($1.4B division of $14B HCL Technologies), orchestrating product strategy across 4 continents. Previously shipped at a Stanford X-backed startup, built a PM function from scratch at an Ed-Tech company, and transformed operations at Fortune 500 banks.
B.Tech in Computer Science, GMAT 770 (99th percentile — highest achievable score).
My edge is that I sit at an intersection almost no one occupies: I can file a USPTO patent for an AI engine, present its business case to a Fortune 500 CTO, ship 50+ releases across globally distributed teams, AND bootstrap a $100K ARR spice business with a 30-member all-women workforce — simultaneously.
What I care about most:
├── How do you turn AI capability into $30M+ pipeline (not just a demo)?
├── When should the AI decide, and when should a human stay in the loop?
├── How do you build products that work for a Top 3 German Bank AND a D2C startup?
├── What does product-market fit look like when the model is probabilistic?
└── How do you scale across 4 continents without losing velocity?
| $100M+ | $75M+ | 50+ | 4 | 770 | 7,000+ |
|---|---|---|---|---|---|
| ARR Managed | Leakage Prevented | Releases Shipped | USPTO Patents | GMAT Score | Children Educated |
| AI/ML & Data Suite | F500 US Bank (550% ROI) | 6 Platforms Launched | AI Engines Invented | 99th Percentile | Sir Syed Society |
1. The patent test. If your AI feature can't be described as a novel system — with clear inputs, processing logic, and differentiated outputs — it's a wrapper, not a product. I've filed 4 patents because I build engines, not integrations.
2. Fallbacks are the product. Every AI feature has a failure mode. I design the fallback first, not last. A rule engine isn't a consolation prize — it's what keeps your product running when the model is unavailable, slow, or wrong.
3. Metrics before magic. "We added AI" is not a roadmap item. "$30M pipeline from a new AI engine that reduced manual compliance review from 4 hours to 12 minutes" is. Every AI feature I ship starts with the business case, not the model card.
4. Scale is a design choice, not a consequence. Building for a Top 3 German bank (€20M+ daily transactions) forces different architectural decisions than building for a startup. I've done both — simultaneously.
5. Ship, then optimize. Clarion went from idea to 30 users in 15 days. Yum Spices went from $5K to $100K ARR. Velocity isn't about cutting corners — it's about knowing which corners don't exist.
current:
- How agentic AI changes enterprise workflows (built it: Glaux's autonomous process mining)
- LLM evaluation frameworks that correlate with business value, not just accuracy
- When to use RAG vs fine-tuning vs prompt engineering (shipped all three)
- AI observability as a product category, not a feature (building Strix for this)
exploring:
- Multi-agent orchestration for enterprise compliance (Glaux + regulatory engine)
- AI-native product management — specs that machines can verify (Clarion's MCP bridge)
- Sustainable AI: data center energy optimization (shipping DCEM at HCL)
- The gap between "AI can do this" and "users trust AI to do this"
believe:
- The best AI PMs can read a model card AND present to a Fortune 500 CTO
- A working prototype changes the conversation faster than a 50-slide deck
- Enterprise scale and startup speed are not opposites — I prove this daily
- The PM who understands the data pipeline has an unfair advantage
Clarion — AI-Native Product ManagementA full-loop PM platform I designed and built that demonstrates what serious AI product work looks like when you close every gap in the workflow. Every architectural decision reflects real PM thinking — where AI takes the wheel, where humans must stay in the loop, and how you make product discovery fast without making it shallow. Why it showcases AI PM thinking:
|
|
What's inside the AI engine:
| Module | What it does | Why it matters |
|---|---|---|
| Signal Collector | Ingests from 20+ sources (Zendesk, Intercom, Gong, Slack, CSV, webhooks) | No PM tool does this — signals are scattered across 10+ tools |
| Intelligence Engine | Classification, sentiment, theme extraction, embeddings (pgvector) | Raw feedback becomes structured, searchable, prioritized |
| Opportunity Canvas | "What should we build?" with RAG over your entire signal corpus | Evidence-based prioritization, not gut feeling |
| Spec Studio | Dual-format generation — human PRDs + machine SPEC.md | Coding agents (Cursor, Claude Code) can consume specs directly |
| Agent Bridge | MCP server exposing 7 tools for AI coding assistants | The bridge between product intent and code execution |
| Validation Engine | AI-drafted email outreach, surveys, concept testing | Prove demand before building — the hardest PM discipline |
| Outcomes Tracker | Feature-to-signal linking, ROI measurement, scoring calibration | Close the loop — did what we built actually solve the problem? |
|
A full-stack observability platform I'm designing that sits at the convergence of two massive markets — APM ($5.1B) and AI monitoring ($7.29B, 40.5% CAGR). Every architectural decision reflects a core insight: LLM workloads need observability that's built native, not bolted on. Why it showcases systems-level PM thinking:
|
→ Strix on GitHub
|
-00ADD8?style=flat-square&logo=go&logoColor=white){kind=icon}
What's inside the observability stack:
| Layer | What it does | Why it matters |
|---|---|---|
| Metrics (TimescaleDB) | Time-series storage with continuous aggregates, multi-tier rollup (raw → 1m → 5m → 1h → 1d) | Hot queries need sub-second response; cold analytics need cheap storage — one DB can't do both |
| Traces (ClickHouse) | Distributed tracing with flame graphs, service maps, tail-based sampling | Columnar OLAP gives 200x cheaper storage than row-based alternatives |
| LLM Prompt Registry | Versioned prompts with LCS diff viewer, playground for testing, annotation queues | Prompt changes break production silently — version control makes regressions detectable |
| LLM Evaluation | 6 automated scorers (factuality, relevance, toxicity, PII, format, injection detection) | "Is the AI output good?" needs a framework, not a vibe check |
| RAG Analytics | Embedding drift detection, retrieval relevance scoring, chunk quality analysis | RAG pipelines degrade silently — you need observability into the retrieval layer, not just the generation |
| Model A/B Comparison | Side-by-side evaluation with weighted verdict across cost/latency/quality | Switching models is a product decision, not an engineering one — PMs need data, not opinions |
| Agent Monitoring | Loop detection, tool-use tracking, multi-step workflow visualization | Agentic AI can get stuck in infinite loops — you need circuit breakers with visibility |
React 18 Express.js MongoDB Razorpay Socket.IO
Python FastAPI React Flow PostgreSQL Claude AI
React 18 Supabase Claude AI GPT-4 WebSocket
Core AI / ML Depth
Product & Strategy
Data & Experimentation
Technical (I Build)
Domain Depth
2024 ━━ GROUP MANAGER (ASSOC. DIRECTOR) ━ AI R&D Lead ━ HCL Software
4 Patents · $25M Pipeline · Stevie Award · Google/Anthropic/Cisco · 4 Continents
2022 ━━ SENIOR MANAGER — PM ━ HCL Software ($1.4B)
$100M+ ARR · 50+ Releases · World's 1st Business Process AI · F500 Clients
2020 ━━ DEPUTY → MANAGER — PM ━ HCL Software
Google Cloud Marketplace · 14 Verticals · $3.5M Net New · Top 3 German Bank
2019 ━━ SENIOR PM ━ Centilytics (Stanford X)
5 PMs · $500K+ Savings Month 1 · $5M Pipeline
2018 ━━ SR. CONSULTANT PM ━ Managineers India
Built PM Function from 0 · 8-Member Team · 2x Revenue Growth
2016 ━━ SENIOR BUSINESS ANALYST ━ HCL Technologies ($14B)
Fast-Tracked Top 5% · European Bank $800K/Year Savings
|
MyXalytics Data Analytics & Reporting |
iControl Business Flow Observability |
IntelliOps Full-Stack AIOps Suite |
FinOps Cloud Cost Management |
iObserve Full-Stack Observability |
DCEM Data Center Energy Mgmt |
Telco Obs. 5G Network Observability |
| Year | Award | Organization |
|---|---|---|
| 2025 | Gold Stevie Award — Innovation in Technology | International Business Awards |
| 2025 | 2Good Award — Exceptional Performance | HCL Software |
| 2024 | AI Innovation Award — Process Mining & Regulatory AI | HCL Software |
| 2024 | Silver Stevie Award — AI Product Excellence | International Business Awards |
| 2024 | 4 USPTO Patents Filed | U.S. Patent Office |
| 2022-23 | Distinguished Performance (2x Consecutive) | HCL Software |
| 2020 | GMAT 770 (99th Percentile) | GMAC |
"I don't just manage products — I build them. From $100M ARR enterprise suites to $100K ARR bootstrapped ventures. The best products emerge when you understand both the business problem and the technology deeply enough to make hard tradeoffs."
Open to AI PM leadership roles across Enterprise SaaS · AIOps · Observability · FinTech · Regulated Industries
