Hi, I'm
Pratheep J R
Product Engineering Leader | AI Platform & Agentic Systems | eCommerce, App & Web Platforms at Scale
I lead engineering teams building large-scale, customer-facing platforms — currently Associate Director of Engineering at Best Buy, running pricing & deals engineering for a multi-billion dollar revenue business. After 16+ years across retail, supply-chain, and finance, my focus now is AI-first: designing agentic architectures, transforming developer experience with AI agents across the SDLC, and scaling teams that hold up under Black Friday load. Outside of work, I'm a father.
Experience
16+ years leading engineering teams and platforms.
Associate Director of Engineering, Pricing & Deals · Best Buy
2025 – Present- Own engineering strategy for pricing & deals across a multi-billion dollar revenue business — scalable discounting, personalization, and high-availability Web/App experiences.
- Drove AI-agent adoption as a peer across planning, build, test, and maintenance — 65% more developer productivity, 50% faster feature turn time.
- Lifted revenue participation by $200M by improving offer placement, personalizing promotions, and enabling the business to curate offers directly.
- Increased Return on Ad Spend (RoAS) by ~$90M by unifying the deals experience across channels.
- Championed cross-platform UI libraries and led the technology roadmap for personalized, Black Friday/Cyber Monday-scale deals and pricing across a polyglot cloud stack (AWS, GCP).
- Scaled and led 60+ engineers and leaders globally, enabling autonomous delivery through clear ownership and aligned team charters.
- Built on a stack spanning React, React Native, and Next.js (SSR), with Sentry, SpeedCurve, and similar tooling for monitoring and performance.
Associate Director of Engineering, Self-Service, Mobile, App & Web · Best Buy
2023 – 2024- Launched an AI-driven customer support experience on .com powered by a machine learning model trained on support data to derive customer intent from natural language, bridging chat, IVR, and .com through a shared, context-driven system — cutting operational costs by millions annually.
- Delivered sub-100ms personalization services sustaining peak performance through Black Friday load; shipped Best Buy's highest-downloaded app during BF 2024.
- Led global web & app teams managing $10M+ in investments to ship high-impact features across React and React Native.
- Unified web and app development workflows via shared infrastructure and component reuse, cutting dev time and cost by 60%.
- Built on a stack spanning Python, React, React Native, Node.js, Express, Java, and GraphQL.
Senior Engineering Manager, Geek Squad Services · Best Buy
2020 – 2023- Designed end-to-end eCommerce for parts — merchandising, catalog, orders, logistics, and reverse supply chain — with modern APIs and ERP integration.
- Established stream-aligned teams shipping 1,000+ deployments/month via a unified platform architecture.
- Migrated analytics and reporting to GCP, cutting infrastructure costs 50% and deployment costs 60%.
Technology / Solutions Architect · Best Buy (Geek Squad Services) / Accenture
2016 – 2020- Applied Domain-Driven Design to architect loosely coupled, scalable eCommerce supply-chain platforms.
- Built CQRS-based event-driven microservices; delivered Best Buy's first automated ink replenishment solution for HP, Epson, and Canon.
- Led the Apple ASP program, balancing build-vs-buy decisions against long-term enterprise architecture.
Community & Board
Board Member · BestPrep
March 2026 – PresentBestPrep is a Minnesota nonprofit (founded 1976) that builds business, career, and financial literacy skills for students in grades 4–12 through hands-on programs — reaching over 1.8M students and educators to date.
BELA Representative, Cloud Coach Program · BestPrep, Best Buy
APAERG Co-Lead, VAJRA Lead · Accenture
Selected Work & Initiatives
AI platform and engineering leadership work I've been part of recently, partnered with peers across the org.
Deals & Pricing Performance Overhaul
Lifted the performance score for deals and pricing experience components from 30% 'good' to 90% 'good' — using component-level (L2) caching, SSR paired with CSR, Intersection Observer-based lazy loading, layout optimization, and bundle-size reduction.
AI-Driven Customer Support Experience
Launched an AI-driven customer support experience on .com, powered by a machine learning model trained on customer support data to derive customer intent directly from natural language. Bridged the experience across chat, IVR, and .com through a shared, context-driven system.
Agentic Architecture & Developer Platform
Partnered with peers to design a layered agentic architecture — Core, Orchestration, Doer, and Validation agents mapped to a 'brain model' — and led its evolution into an enterprise developer agent platform: local and GCP agent registration, secure access to enterprise agents like an org knowledge-resolution agent, and a clear framework for when to use Agent-to-Agent (A2A) protocol versus MCP. Also evaluated tooling such as LangChain/LangSmith for fit.
AI Engineering Platform
Partnered with peers on foundational AI engineering infrastructure — an internal AI-first context engine giving agents and tools grounded, up-to-date context about the business, and a 'software factory' model for an AI platform designed to multiply engineering throughput across teams.
Agentic Ways of Working
Partnering with peers and enterprise leaders on the AI ways-of-working and org/taxonomy/ontology structure organizations need to realize value as agentic software development matures.
Blog
Writing about engineering, architecture, and the things I learn along the way.
Agents Don't Change Your Org Chart — They Change What It Can Hold
On applying Team Topologies to an agentic engineering org: diagnosing duplicated capability spread across too many teams, redesigning around durable value-stream-aligned squads, and the real insight — agents don't reshape team boundaries, they change what a squad's own workflow can absorb before work has to hand off to another team.
Why We Modeled Our Knowledge Graph as an Agent, Not an API
A layered agent architecture — input, reasoning, coordination, execution, judgment, memory, control plane — and the case for treating org/domain knowledge resolution as an autonomous, reconciling agent rather than a static lookup API. Plus a sharper distinction than it first seems: why feedback loops belong on stateful, multi-turn A2A rather than stateless MCP tool calls.
The Brain as an Operating Model for Agentic Enterprises
Framing an enterprise as an adaptive system — Sense, Triage, Protect or Deliberate, Coordinate, Act, Learn — and mapping it to a brain: reflexive/protective agents for incidents and fraud versus deliberative agents for strategy and architecture, with memory, threat-weighting, and execution each playing a distinct role.
Ask Me Anything
Chat with my AI agent to learn about my career, projects, and blog posts.
Start a conversation! Try asking:
- "What projects has Pratheep built?"
- "What are his skills?"
- "What has he written about?"