Hi, I'm Shubham.
I build intelligent platforms.
Software Engineer 3 @ Walmart Labs. Specializing in AI/LLM integrations, distributed backends, and data platforms processing hundreds of millions of events.
Career Journey
Software Engineer 3
Walmart Labs, CA (E-Commerce)
- Replenishment & AI Integration: Designed a data platform processing 550M+ SKUs with predictive analytics and ML models to optimize inventory forecasting.
- Database LLM Chatbot Integration: Developed a natural language database chatbot using Google Vertex AI and MSSQL. Transformed user queries into optimized BigQuery SQL dynamically.
- Advanced API Development: Conducted in-depth analysis and built scalable API-driven microservices using Java Spring Boot, Apache Spark, Airflow, and GCP.
Software Engineer 2 - Big Data
Choice Hotels, AZ (Hospitality)
- Data Analytics Platform Enhancement: Optimized live data streams and analytics using AWS (Airflow, EC2, EMR, Athena).
- Automated legacy system upgrades for extreme reliability and high-availability.
Software Engineer
Early Warning, AZ (Fintech / Zelle)
- Backend Enhancement: Enhanced the Zelle Payment App backend with Java Spring Boot, Docker, and REST APIs to securely process high-value transactions.
- Cloud & AI Integration: Supported AWS cloud upgrades and implemented a streamlined notification service enabling AI-driven features.
Masters of Computer Science
Arizona State University (GPA: 3.7/4.0)
- Specialization in Data Mining and Distributed Database Systems.
- Served as CSE Tutor for Data Structures and Algorithms.
Featured Projects
Production-minded systems for building, serving, and evaluating LLM applications — plus the backends that carry them.
go-llm-gateway
Production API gateway for LLM traffic — rate limiting, cost tracking, and automatic model failover. Zero third-party dependencies.
vertex-ai-rag-pipeline
End-to-end RAG: Vertex AI text-embedding-004 for embeddings, BigQuery Vector Search for retrieval, and Gemini 1.5 Pro for grounded generation.
llm-eval-harness
Evaluation framework for LLM outputs — faithfulness scoring, correctness checks, LLM-as-judge, and a batch runner to keep quality in the loop.
omnimart-review-agent
Agentic application composing the gateway, RAG pipeline, and eval harness — demonstrating agent loops, tool use, model routing across Claude tiers, and eval-in-the-loop.
omnimart-ratings-reviews
Enterprise-grade Ratings & Reviews service in Go — hexagonal architecture, idempotent writes, cursor pagination, and split site-facing / internal APIs.
bq-genai-poc
Natural-language-to-SQL chatbot — Gemini Pro + LangChain turn plain-English questions into optimized BigQuery SQL. POC of the production system built at Walmart Labs.
Technical Arsenal
AI & Machine Learning
LLM Technologies, Prompt Engineering, Model Fine-tuning, Agent Architectures, Vector Databases, Knowledge Graphs, TensorFlow, PyTorch, Scikit-learn.