Engineering Roadmap

What I'm planning to build next.

These are planned engineering projects — not yet built, not yet started. Listed here to communicate my engineering direction and what I'm working toward by 2027.

Transparency note: Everything on this page is PLANNED or EXPLORING — not completed work. I believe in being clear about the distinction between what I've shipped and what I aspire to build.

01Planned

Multi-Vendor Commerce Platform

Target: 2026–2027

MarketplaceMulti-VendorCommerceSystem Design

A production-grade multi-vendor marketplace with vendor onboarding, dashboards, product catalog, SKU management, inventory, orders, commissions, and analytics. The goal is to understand and implement the full complexity of a real marketplace architecture.

Planned Features

·Vendor onboarding, KYC, and dashboard
·Product variants and SKU management per vendor
·Order routing and vendor-level fulfillment
·Commission calculation and payout system
·Admin controls: approval, moderation, analytics
·Vendor isolation at the data and API layer

System Areas

Vendor data isolationInventory managementOrder routingCommission engineReturns & refundsMarketplace admin

02Planned

AI Commerce Engine

Target: 2027

AICommerceRAGLLMVector SearchSemantic Search

An AI-powered commerce layer with natural-language product search, personalized recommendations, review summarization, and RAG-based product knowledge. Exploring how LLMs and vector search can improve commerce discovery and conversion.

Planned Features

·Natural language product search (semantic + hybrid)
·AI-powered product recommendations
·Review summarization via LLM
·RAG-based product knowledge base
·AI shopping assistant
·Product comparison engine

System Areas

Semantic searchVector embeddingsRAG pipelineLLM integrationRecommendation engineReview intelligence

Context

Why these projects?

My current work — VedaGlow and GharBazaar — covers the core of single-vendor eCommerce and marketplace systems. The next natural step is to understand the complexity that emerges when you scale: multiple vendors, more sophisticated inventory, commission systems, and AI-driven discovery.

The Multi-Vendor Commerce Platform is about understanding marketplace architecture at a deeper level — vendor isolation, order routing, financial reconciliation. These are the problems that platforms like Meesho, Myntra, and Amazon Seller Central solve at scale. I want to understand these patterns from the inside.

The AI Commerce Engine is about the intersection of LLMs and commerce — specifically how natural language search, semantic recommendations, and RAG-based product knowledge can improve discovery and conversion. This is the direction commerce engineering is moving in the next 3–5 years.

Both projects are PLANNED. I will update this page as they move to IN DEVELOPMENT and eventually to completed work.

Current Focus

What's actually in progress right now.

Live

VedaGlow

D2C eCommerce platform — shipped and live. Ongoing iteration and improvement.

Case Study

In Development

GharBazaar

Property marketplace with university-based student housing discovery.

Case Study