Headline

AI Product Manager @Coolboots | Shipping production AI for career products serving 2L+ users | Hands-on with model selection, prompt pipelines, voice agents | Outcome-driven PM

About

Most PMs in AI write specs. Some PMs in AI also pick the model, design the prompts, and rebuild the pipeline when it doesn't work. I'm the second kind. Currently AI Product Manager at CoolBoots, owning monetization and growth funnels across career products serving 2L+ users. The work I'm most proud of in the last 6 months: → Co-led the migration of our Gold Resume product from a GPT-4-class model to an optimized Qwen 2.7B + prompt system — cut cost per resume >70%, contributed to ₹1Cr+ revenue through funnel optimization → Shipped Job Alerts 0→1 with semantic matching and relevance scoring, driving ₹1.3Cr revenue in first month → Built an AI voice agent that replaced high-CAC telesales and lifted trial-to-paid conversion 40% → Stabilized our LinkedIn Makeover product by identifying inconsistency in AI outputs and standardizing the generation pipeline — cut refunds by ₹2L/month Before this: Product Operations Associate at Coursera (co-owned the chargeback dispute tool that recovered $2M+ annually and cut response time 76%). Before that: Data Analyst at Enstoa, building SQL + Power BI dashboards. The arc, honestly: I started in data, moved to product ops, and ended up doing AI PM work because I kept moving toward where the AI was being shipped. I write about three things: → Decisions early-career AI PMs face that they don't get taught (model selection, cost vs quality, when to ship vs polish) → What "outcome over output" actually looks like in AI products → Receipts from the AI features I ship — what worked, what failed If you're a senior PM or hiring manager looking at early-career AI PMs — or another PM trying to figure out the same questions I am — let's talk.

Location

New Delhi

AI stack

aistudio

claude code

cursor

elevenlabs

openai

OpenRouter

Work experience

AI Product Manager at CoolBoots, Nov 2025 - Present · 1 yr, Gurugram, Haryana, India, • Owned monetization and growth funnels across AI-driven career products serving 2L+ users, defining AI product strategy across revenue, conversion, and unit economics. • Launched 2 zero-to-one offerings — Visibility+ (₹30K) and Career+ (₹50K) — delivering end-to-end LinkedIn growth and career assistance through multi-agent systems for onboarding, outreach, engagement, and job applications. • Launched a RAG + vector-based Job Alerts engine, achieving 85% match relevancy, onboarding 2,700+ users, and generating ₹54L MRR in the first month. The same engine contributed to 30% of interviews generated for Career+ customers. • Built AI-led growth systems including a LinkedIn outreach agent and AI voice agent, combined with redesigned onboarding and messaging flows, increasing lead-to-trial conversion by 40%. • Designed a 6-agent LinkedIn content orchestration system covering input structuring, content planning, post/carousel/infographic generation, model selection, and automated quality checks; scaled to 10,000 users. • Reduced resume generation costs by 70%+ by migrating to an optimized Qwen 2.7B + prompt architecture and eliminating DOCX-to-PDF dependency through OpenAI Vision. • Designed a lead and job-fit qualification system using TypeSafe AI’s Jev model, achieving 98% decision accuracy and automating high-volume fit classification. • Leading an MCP-based autonomous job application system across Naukri, Indeed, and LinkedIn Easy Apply to automate applications for Career+ users. • Negotiated AI vendor pricing, reducing per-seat costs by 89% and 20%, while securing free trials and maintaining 8+ month vendor partnerships.

Product Operations Associate (Product Services) at Coursera, Nov 2024 - Nov 2025 · 1 yr 1 mo, Gurugram, Haryana, India, Product Operations Associate at Coursera, working across Product Services. I owned QA workflows, partnered with Finance on dispute resolution, and worked with Enterprise + Engineering on tailored client solutions. Selected wins: → Chargeback dispute resolution tool — Co-owned the design and pilot of a software-based dispute resolution tool with Finance. Outcome: $2M+ annual reduction in refund losses, 76% cut in response/evidence prep time (17 minutes → 4 minutes), and $130K in revenue recovery during the pilot phase. → AI feedback agent on Vertex AI — Led the development of an AI feedback agent using Vertex AI + Gemini API. Fine-tuned an LLM on 100K+ learner feedback records to convert qualitative data into actionable insights. Outcome: improved predictive accuracy and enabled 40% faster partner response cycles. → Course QA at scale — Drove end-to-end QA for new and existing courses; established compliance workflows that reduced course-related errors by 25% and improved partner satisfaction. → Enterprise deal support — Partnered with Enterprise and Engineering teams to deliver tailored data solutions for client proposals and optimize localization workflows (machine translation + AI dubbing). Outcome: 30% improvement in global rollout efficiency, contributed to multi-million-dollar deals with private and government institutions.

Analyst at Enstoa, Oct 2023 - Jun 2024 · 9 mos, New Delhi, Delhi, India, 1.5 years total. The foundation of the data-thinking I now apply as an AI PM. What I owned: → SQL + Power BI dashboards that communicated business performance, daily operations, and client-specific system metrics — feedback loop with Product Managers and Client Success Teams led to a 20% decrease in technical issues across projects. → Support ticket analytics using Zendesk data — cut data-related ticket SLA by 45%. → Cross-functional work with engineering and business teams (Scrum/Agile) — implemented data-driven monitoring that improved end-customer experience and stakeholder engagement (30% lift in stakeholder engagement via clearer communication of technical concepts to non-technical audiences). What I learned at Enstoa that I use every day in AI PM work: if you can't measure it, you can't ship it. Every AI feature I touch at CoolBoots gets instrumented first, built second.

Junior Technology Analyst (Junior Analyst) at Enstoa, Jan 2023 - Sep 2023 · 9 mos, New York, United States

Data handling intern at Marquee Equity, May 2022 - Jul 2022 · 3 mos, Remote, 1.) Creating backup of pre-sales data . 2.) Conducting analysis on regular basis to understand the variation in data. 3.) Sorting of data .

Operation intern at TravClan, Mar 2022 - May 2022 · 3 mos, New Delhi, Delhi, India, I use to solve queries of travel agents in the form of quotation with the help of analytical data . Working at backend of Maldives.Travclan portal updating Data of multiple resorts by using (Google sheets, Ms-Excel)

Education

Bachelor of Technology - BTech, Civil Engineering, Dr Akhilesh Das Gupta Institute of Professional Studies

Public links

substack.com: https://substack.com/@darp1