I'm an AI product leader at UPL, building enterprise generative AI products: LLM applications, AI agents, RAG systems and AI governance. Before UPL, my work at HireRight included multimodal LLM document verification, GenAI assistants covering 4,000+ courts across 2,600 U.S. jurisdictions, and automation that saves $125K+ a year.
I work where product strategy, LLM engineering and responsible AI meet. I write PRDs that engineers trust, make the prioritization calls the business needs, and govern AI so it is safe to ship. Outside work, I teach AI through a book, a YouTube channel, articles and open-source projects anyone can run.
MIT Pune (Computer Science) · IIM Mumbai (MBA) · 8.5+ years in B2B SaaS, ML and data platforms · Currently UPL · Previously HireRight, DevOpsLabs, Sterlite Technologies, Deloitte, Infosys
| 💰 $125K+ | ⏱️ 80% | 🎯 95% | 🏛️ 4,000+ |
|---|---|---|---|
| Annual savings from a multimodal LLM | Faster document verification (TAT) | Accuracy on AI-powered KYC | Courts covered by the scraper platform |
| 🤖 5 FTEs | 📈 30% | ⚡ 20% | 🗓️ 8.5+ yrs |
| Workload automated by a GenAI chatbot | CSAT improvement | Faster response time | B2B SaaS · ML · data platforms |
|
The Engineer's Complete Guide to Modern AI Systems. A 537-page field manual (33 chapters, 12 appendices, 100K+ words) on building, deploying, scaling and running real-world AI systems, from tokens and embeddings to production AI platforms. Free to read.
|
New videos every day. 30-second explainers ("What is RAG?", "What is a vector database?") and 2-minute "How it works" stories (Uber, Google Maps, WhatsApp, Google Search). Made with my own AI video pipeline. |
|
|
- Start with the metric, not the model. Every AI feature has a business number it must move: savings, turnaround time, accuracy or CSAT.
- Evaluate before you scale. Golden datasets, red-teaming and LLM-as-judge come before launch, not after the first incident.
- Govern by design. Privacy, compliance (GDPR, OWASP LLM Top 10) and human-in-the-loop are product requirements.
- Agents earn autonomy. Start with workflows. Give an agent more freedom only when tests and guardrails show it can handle it.
- Ship, then teach. I build each idea, then share how it works so others can build it too.
|
⚙️ AI Automation Workflows · new Three end-to-end automations built with Claude:
|
A local-first voice assistant to think out loud with. It shows its reasoning as it goes and has four layers of memory across sessions. Model routing switches to a backup model if one fails. Wake word, voice detection, speech-to-text and text-to-speech all run locally.
|
|
Turns a folder of photos and music into a cinematic film from a plain-English brief. An LLM plans the edit and a separate, predictable renderer produces it, so the two jobs are kept strictly apart.
|
🏢 Enterprise agent suites
|
|
🔐 Governance Agents: 4 agents check enterprise tool requests against FCA, PRA, UK GDPR and SMCR rules and return a governance report in under 60 seconds. |
🛡️ AI Agent Evaluation Platform: red-teams AI agents across 12 attack categories aligned with the OWASP LLM Top 10 (jailbreaks, prompt injection, hallucination, data leakage). |
|
🕵️ Sentinel AI: a security dashboard for LLMs, RAG pipelines and agents, with a remediation copilot. Mapped to SOC 2, GDPR, HIPAA and ISO. |
✍️ PromptPilot AI: a Chrome extension that scores prompts on 7 quality dimensions and cuts 20–40% of tokens across 8 AI platforms. |
📦 More enterprise AI products
- MoEngage AI Agents: campaign analyst, churn-risk prediction, A/B testing and recommendation agents
- PulseIQ: a 0–100 client health score, 30-day churn forecast and AI copilot chat
- LLM Projects 2.0: 9 production AI apps (Text-to-SQL RAG, KYC automation, meeting summarizer), each with a PM case study
- Voice Agents: 6 voice-first apps, including Jira by voice, HR document Q&A and market briefings
- Agentic Marketplace: product design and prototype for an AI agent marketplace
- AI Interview Coach: LLM-generated PM/TPM questions and answer evaluation
- 🥇 Sense of Ownership Award ×3, HireRight (2023, 2024, 2024)
- ⭐ Star Performer of the Month, Sterlite Technologies (2021)
- 🏁 Business case competition finalist at YES BANK, Flipkart, HUL and Reckitt
Open to: AI product leadership roles · Speaking & podcasts · AI strategy advisory · Collaborations
🏢 UPL · 📍 Bengaluru, India
AI Product Leader · AI Product Manager · Head of AI Product · Generative AI · LLM Products · AI Agents · Agentic AI · RAG · LLM Evaluation · AI Governance · Responsible AI · Enterprise AI · B2B SaaS · Claude · OpenAI · LangChain · CrewAI · AI Educator · AI Author · AI YouTube