Q: Tell me about yourself / walk me through your background.
A: I am a product-marketing and go-to-market leader focused on making technically complex products understandable, credible, and commercially useful. I began at McKinsey, where I learned to connect customer research, pricing, and commercialization to measurable business decisions. At D. E. Shaw, I applied that discipline in a highly regulated environment, rebuilding institutional marketing around evidence, claims governance, and value realization rather than campaign volume.
At Microsoft, my scope expanded across enterprise segmentation, lifecycle growth, cloud migration, augmented reality, and AI-assisted advertising. That work taught me that the buyer promise, technical proof, field enablement, and compliance model have to be designed together. During my MBA experience at Rakuten, I worked on localized AI adoption in Japan, including the decision to delay a recommendation launch until language, relevance, latency, and data handling met the market bar. At AWS, I have focused on AI product marketing and strategic GTM, including conversational-shopping decision support and evidence-gated investment cases.
The through-line is that I do not treat messaging as decoration around a product. I define the customer decision, build the proof and launch system behind it, align Product, Sales, Engineering, Legal, and Finance, and measure whether positioning changes adoption, pipeline, conversion, retention, or investment. I am strongest in roles where an important technology needs both a sharper market story and a more credible path to scale.
AI products do not reach a market because the model is impressive. They reach it when a customer can recognize the decision the product improves, trust the evidence behind the promise, and see a credible path from first use to economic value.
That is the work represented here. Across conversational commerce, spatial shopping, recommendations, advertising, cloud migration, investment products, and retail pricing, I have treated positioning, launch readiness, adoption, compliance, and value realization as one system.
Resume · LinkedIn · shiv-ai-pmm@umich.edu
My product-marketing decisions begin before the campaign. I define the customer decision, determine which facts must remain authoritative, identify what could make the promise unsafe or unbelievable, and build the measurement needed to earn broader distribution. The message is the visible edge of that operating model.
The seven projects form a progression from product meaning to market scale.
In Alexa Shopping: conversational decision support, roughly 100,000 interactions exposed a 40% drop-off when shoppers left the assistant for conventional search. I repositioned the experience from open-ended conversation to a sourced, constraint-aware shortlist that preserved customer confirmation. A 15,000-shopper cohort then connected claim quality, journey continuation, confidence, and adoption to a further $5 million investment decision.
AR shopping: global launch and return reduction applied the same discipline to a different source of doubt: whether a product would fit in the customer's physical space. The proposition shifted from novelty to pre-purchase evidence, joining three market pilots, a governed 3D-asset supply chain, page-performance limits, and return economics. The recorded evidence associated AR engagement with 25% lower returns and 9% higher enabled-SKU sales while preserving the selection limits around the 40% conversion comparison.
For the Japan recommendation-engine launch, I recommended a six-week delay rather than market a translated interface as a Japan-ready product. Respectful language, tokenization, local relevance, latency, evidence, and data handling became release conditions. The work demonstrates a central belief: a launch date is reversible; a breach of customer trust is not.
In generative-AI advertising, trust moved inside the creation experience. More than 500 pages of policy became contextual rules, visible coaching, constrained drafting, joint copy/image inspection, and human escalation. Rejection fell from 18% to 1.8%, creation moved from five-to-seven days to under an hour, and shadow evaluation recorded 96% precision and 91% recall. Compliance was not a review queue after the product; it was part of the value proposition.
Manufacturing cloud migration began with a twelve-to-fourteen-week content process that could not keep pace with a four-week launch. I changed the unit of production from an asset to a buyer decision, brought architects and writers into live working sessions, and governed reuse. The team delivered 20+ decision products in four weeks with zero late technical-rework cycles, moved technical-lead trial conversion from 5% to 15%, shortened the influenced sales cycle from twelve to nine weeks, and avoided $120,000 of planned agency spend.
The quant-fund marketing rebuild shows the same operating principle under regulatory pressure. After roughly a quarter of more than 1,000 reviewed assets required intervention, I rebuilt the system around claims that carried their proof. Ninety percent of 50 paused conversations resumed, $50 million of pipeline reopened, $12 million of new AUM was attributed to the recovery, and review time fell 40%—with pipeline, AUM, and fee revenue kept analytically separate.
In retail pricing intelligence, I rejected a maximum-price promise in favor of portfolio decisions merchants could understand and challenge. SKU economic roles, top-500 item locks, a ±5% weekly cap, evidence cards, controlled testing, and governed overrides translated analytics into an adoptable commercial product. The local publication cycle fell from six weeks to under one, overrides dropped below 5% by month two, and a $50 million opportunity remained explicitly modeled rather than claimed as realized revenue.
Q: How would you position and message a product?
A: I start with the customer decision, not the feature list. On the Alexa Shopping work, review of roughly 100,000 interactions showed that 40% of shoppers dropped out when they had to leave the assistant and return to conventional search. The unmet job was not “have a conversation.” It was “help me reduce a large choice set to a defensible decision without losing control.”
Using a Value Proposition Canvas, I translated that into three layers. The job was to find an eligible product for a specific need. The pains were stale commerce facts, repetitive searching, incomparable specifications, and fear that an assistant might act on an inferred preference. The gains were a small, diverse shortlist, differences explained in the shopper's context, sourced evidence, and explicit confirmation before action.
That produced the positioning: conversational decision support that turns a shopper's constraints into a current, explainable shortlist while leaving the final decision with the customer. The messaging hierarchy then followed:
- Outcome: reach a confident choice with less search and comparison work.
- Defensible capability: constraint-aware ranking and contextual comparison, grounded in authoritative commerce systems rather than generated claims.
- Trust: visible sources, surfaced conflicts, correctable memory, and customer confirmation.
- Proof: a 15,000-shopper cohort, a recorded 2.8% claim-error rate, brand favorability moving from 55% to 70%, and a further $5 million investment decision.
The competitive contrast was not only another assistant. Conventional search returned abundance without synthesis; reviews supplied opinion without guaranteed fit; an unbounded chatbot could sound confident while using stale facts. Our position combined selection, explanation, live evidence, and customer authority.
I aligned Product and Science around decision quality rather than conversation volume, Commerce around authoritative price, availability, delivery, and returns, Brand and Legal around the claim hierarchy, and Finance around an evidence-gated expansion case. The same approach appears elsewhere in this portfolio: AR was positioned as pre-purchase fit evidence rather than novelty, and retail pricing as explainable portfolio decisions rather than maximum-price automation. My rule is that positioning is finished only when the promise, product behavior, proof, and commercial model tell the same story.
I operate where product truth becomes market behavior. My work spans audience and journey research, positioning, category design, launch gates, field and content systems, policy, experimentation, adoption, and value measurement. The common output is not a slogan. It is an organization that can explain what the product does, prove why it deserves trust, and know whether the market has earned the next investment.