¶ Currently

Shipping LLM systems at Walmart scale — and building the next thing.

Walmart Global Tech
Sunnyvale → Mountain View
7+ yrs production
MS CS · Arizona State
01 · About

I lead engineering on an LLM-powered chatbot at Walmart that lets business users query a 550-million-SKU inventory dataset in natural language — zero SQL required. It's built on Vertex AI and BigQuery NL2SQL, and it serves real production traffic every day.

Before that, I shipped distributed data platforms at Choice Hotels, high-value payment backends at Zelle, and ran an ML internship at American Express. The thread through all of it: systems that have to be correct, fast, and still readable two years later.

Right now I'm in Mountain View sketching the foundations of an AI startup — Go microservices, event-driven architecture, single-table NoSQL, the boring decisions that compound. If you're building something similar, let's talk.

02 · Experience
2026 — NOW
Senior Software Engineer
Walmart Global Tech · Sunnyvale, CA
Lead engineering on an LLM chatbot over 550M-SKU inventory. Vertex AI + BigQuery NL2SQL. Spark ML forecasting. Go and Java backends.
2023 — 2026
Software Engineer III
Walmart Global Tech · Sunnyvale, CA
Distributed inventory systems for omnichannel retail. BigQuery analytics, Airflow DAGs, ML model integration into production fulfilment.
2022 — 2023
Big Data Engineer
Choice Hotels · Phoenix, AZ
ETL and batch processing on Hadoop and Spark. Airflow orchestration, EMR cluster ops, data quality across reservation systems.
2021 — 2022
Software Engineer
Zelle · Early Warning Services
High-value payment backend. Java, Spring Boot, REST APIs over PostgreSQL. Strict latency and correctness constraints.
2020
Machine Learning Intern
American Express
Predictive models for transaction risk signals. Python, TensorFlow, scikit-learn on internal data platform.
03 · Selected work

For an interactive architecture diagram, a live semantic-search demo, and a deep-dive essay, visit the Lab ↗.

go-llm-gateway
Production API gateway for LLM requests — per-IP rate limiting, atomic cost tracking, structured logging, primary→fallback failover. Zero external deps.
GoLLM InfraRate LimitingObservability
2026
vertex-ai-rag-pipeline
End-to-end RAG pipeline on Vertex AI. BigQuery Vector Search retrieval, LangChain orchestration, grounding hooks, cost-aware batch processing.
PythonVertex AIRAGLangChain
2026
llm-eval-harness
LLM evaluation framework — faithfulness scoring, correctness vs ground truth, LLM-as-judge, batch runner, CSV/JSON export. CI-ready.
PythonGeminiLLM-as-JudgeEval
2026
omnimart-ratings-reviews
Hexagonal-architecture ratings & reviews service. Dual REST APIs, cursor-based pagination, integration tests, deployed docs.
GoHexagonalPostgresREST
2026
04 · Now
Building
An AI-native review-quality platform — event-driven Go services, single-table DynamoDB, designed for high availability from day one.
Studying
Inference-cost optimisation, agentic eval, retrieval over structured + unstructured corpora.
Reading
Designing Data-Intensive Applications · The Anthropic ToS · Go standard project layout debates.
Stack
Go · Vertex AI · BigQuery · Spark · LangChain · DynamoDB · GCP · AWS
05 · Connect

If you're building what comes next — let's talk.

Most interesting to me: founding-team conversations, AI-native infrastructure, LLM platforms, agentic systems. I respond to every thoughtful message.

shubhambakre@gmail.com Resume