An AI MVP that became the flagship.

CPG giants were auditing shelves with clipboards.

Brands spend enormous field-force budgets checking whether their products are actually on shelf, priced right and displayed per planogram, and the data came back weeks late, on paper. Computer vision could answer in minutes what an auditor answered in days.

The hard part wasn't the model. It was turning a promising demo into a product that enterprise CPG quality teams would bet their KPIs on.

From demo to deployable, the unglamorous work of enterprise AI.

• Led MVP → PMF for ShelfWatch, shaping the product around the metrics CPG sales leadership already reported on, share of shelf, OSA, compliance, not around model outputs.
• Cut AI model onboarding time by partnering directly with Data Science on the SKU-training pipeline, the single biggest blocker to enterprise deployment cycles for large CPG contracts.
• Ran an 11-person cross-functional pod (engineering, data science, design, QA) and shipped two additional product lines, expanding the company's addressable market.
• Closed the loop with field reality: usability testing with actual field reps in actual stores, because a CV product lives or dies on blurry photos taken in bad lighting.

5× to $1–2M ARR, an enterprise logo wall, and an award.

ShelfWatch became ParallelDots' flagship product, scaling 5× to $1–2M ARR with Unilever, P&G, ITC and Nestlé among the enterprise customers. The company won Nasscom Emerge 50 (2020) as Best Retail Tech Startup on the back of it.