AI-First Enterprise Commerce Architect
I design the systems behind large B2B, B2C and B2B2C commerce platforms — SAP Commerce, S/4HANA and CPI integration, performance and security — and put grounded, explainable AI on top of them.
The client-deployed RAG + multi-agent assistant is private. The public Lab shows the layer underneath it: grounding, validation and explainability — mostly deterministic, on purpose.
Each step widened the scope: from building commerce features, to owning data models and performance, to designing integrations and leading delivery — and now applying AI on that foundation.
Technology choices follow the business question. Each layer below earns its place by changing an outcome — and each one has evidence in the Lab.
Select a node — commerce core, integration, enterprise systems and the AI layer around them.
Start from the revenue, cost or service problem — not the platform.
B2B, B2C and B2B2C models on SAP Commerce, kept close to out-of-the-box capability.
Data models and boundaries that stay maintainable as customisation grows.
S/4HANA, Service Cloud, CPI, logistics and messaging as designed, supportable flows.
Measure first: indexing, caching, pagination, query shape. See the benchmarks.
Authorisation at the right layer, every time. See the model.
Rules and decision engines with visible thresholds. See an example.
Grounded, validated and explainable — with a human able to override.
The same foundation — Commerce, Service Cloud, integration — now gets retrieval, agents and guardrails. Each item is labelled by its real status, so nothing experimental is presented as production.
An end-to-end enterprise chatbot answering across product, order and service context from Commerce and Service Cloud. Deployed for one client and being adapted for another.
Built end-to-end across frontend, backend, AI integration and orchestration. Client and system details are generalised for confidentiality.
Decision support that moves from a headline metric to its root cause — retrieval, reasoning and forecasting agents with drill-down — without hopping between reports.
A BTP middle layer connecting Commerce and Service Cloud UIs — agent availability, assignment, conversation history and routing.
Reads purchase history to find delayed reorders, estimates revenue and churn impact, and recommends no action, a nudge or an incentive under configurable rules.
Analyses production logs, matches known issues to a knowledge base and recommends or triggers predefined remediation; new issues go to documentation research with confidence scores.
A product-page assistant that reads intent, documents and media to answer questions, suggest alternatives and bundles, and track purchase-confidence and budget signals.
The Engineering Lab shows the discipline underneath: evidence scoring, citation validation, factuality checks and explainable ranking — mostly deterministic, deliberately.
Portfolio is the story; these are the evidence. Each project has an interactive demo with hard-coded synthetic data, and the full code is on GitHub. Start with the six that best show how I think.
Answers “why did revenue fall?” with ranked, evidenced causes instead of dashboard archaeology.
Hybrid recommender where every suggestion carries the reason it was made.
Structured filters plus a hand-written BM25 ranker over a 124-product catalog.
A plan → search → evaluate → synthesise pipeline that cites evidence or admits it has too little.
Reproducible before/after benchmarks for indexing, caching, pagination and parallelism.
A defensive commerce API: rate limiting, JWT, RBAC, ownership checks and audit logging, mapped to OWASP.
Also in the lab: Churn Prediction API · Transfer Network Analysis · Dynamic Pricing Simulator · Fulfillment Optimizer · Product Discovery Engine · Catalog Quality Auditor · FAQ Generator · SEO Description Generator · Review Summarizer · SEO Score Calculator · Match Intelligence · SAP Commerce Developer Toolkit · SAP Commerce MCP
Four areas with one thread: commerce platform depth, integration and architecture, performance-minded engineering, and applied AI built with guardrails.
B.Tech, Electronics & Electrical Engineering — KIIT University, Bhubaneswar (2018)
SAP Commerce / Hybris · SAP Generative AI · SAP AI Ethics · Generative AI for eCommerce · Prompt Engineering for Web Developers · SAP CX AI Toolkit implementation for SAP Commerce
I’m glad to talk about SAP Commerce architecture, enterprise integration or applied AI. Email is the fastest way to reach me. Based in Kolkata, India.