Hi, my name is

Shubham Chawla

AI & Data Engineer, Google

I am an AI & Data Engineer at Google based in Hyderabad. I specialize in architecting distributed data lakehouses and production agentic AI systems for major banks and enterprise financial workloads.

Contact me!View Projects

Shubham Chawla

AI & Data Engineer @ Google

Hyderabad, India

Big Data Lakehouses & Multi-TB Scalable Pipelines

Agentic AI, GraphRAG & Model Context Protocol

Knowledge Graphs using Graph Databases (Spanner & Neo4j)

Agentic AI Platforms on Google Cloud (ADK)

Application Deployment & Architecture on GCP Infra

Production-Ready Agentic AI for Large Enterprises

Distributed Streaming & OpenTelemetry Tracing

Scroll

Who am I?

I am an AI & Data Engineer at Google with over 10+ years of global experience architecting mission-critical data platforms and distributed systems.

My engineering focuses on building high-throughput Big Data lakehouses and operationalizing production-grade Agentic AI workflows—partnering with major banks and global financial institutions to deliver scalable, secure, and deterministic architectures.

Beyond engineering: Triathlons, Cricket & Chess

Companies Worked With2016 – Present

GoogleCurrent

2022 – Present

Carelon

2021 – 2022

Deloitte

2018 – 2021

Cognizant

2016 – 2018

Featured Projects

A selection of production lakehouses, GraphRAG engines, and open Model Context Protocol tooling.

Slide (8)

AGENTIC AIOpen Source

Equity Portfolio Analyzer Agent (Google ADK)

Multi-source financial agent built on Google ADK fanning out across 4 market data feeds (Yahoo Finance, Twelve Data, NSE/BSE, Tavily) to synthesize an opinionated BUY / HOLD / SELL verdict.

4 Parallel

Data Feeds

~115 Metrics

Data Points

Google ADK

Framework

Google ADKLiteLLMPythonFastAPIFinancial APIsNSE/BSE

MCP TOOLINGOpen Source

Production Database Model Context Protocol (MCP) Suite

Enterprise MCP servers enabling AI agents to safely introspect schemas, generate optimized SQL, and execute database tasks with strict guardrails.

< 140ms

Invocation Latency

100%

Safety Pass Rate

Cloud Run / Docker

Supported Runtimes

Model Context Protocol (MCP)FastAPIGoogle Cloud RunBigQueryCloud SQLDocker

DATA ENGINEERINGOpen Source

GCP Anti-Money Laundering (AML) AI Framework

Enterprise Anti-Money Laundering (AML) transaction compliance pipeline and dashboard deployed on GCP using FastAPI, BigQuery, and Cloud SQL.

Banking AML

Target Domain

BigQuery / SQL

Query Engine

FastAPI / GCP

Backend

Google CloudBigQueryFastAPICloud SQLPythonSQL DDL/DML

AGENTIC AIOpen Source

Agentic AI Observability & Telemetry Framework

Distributed tracing and evaluation framework tracking multi-agent reasoning steps, tool latencies, and token cost telemetry into GCP Cloud Trace.

< 4ms

Trace Overhead

Near Real-Time

Eval Pipeline SLA

Per-Agent Granular

Cost Attribution

OpenTelemetryGoogle Cloud TracePythonVertex AI PipelinesPrometheus

AGENTIC AIOpen Source

Automated Knowledge Graph Construction via LLM

Pipeline leveraging few-shot prompt chaining and graph reconciliation to convert unstructured domain documents into queryable property graphs.

91.8%

Entity Extraction F1

100k+ docs/day

Processing Speed

Strict Schema

Ontology Alignment

Gemini 1.5 ProPythonSpanner GraphPydanticNetworkXGQL

AGENTIC AIOpen Source

SQL Reverse-Engineering & Data Model Agent

AI agent that parses complex enterprise SQL queries, reverse-engineers entity-relationship models, and visualizes schemas in structured JSON.

Structured JSON

Output Format

Streamlit

UI Interface

FastAPI + AI

Architecture

PythonFastAPIStreamlitSQL ParserGoogle Cloud SDKPoetry

DATA ENGINEERINGProduction

High-Throughput CDC & Lakehouse Physical Modeling

Real-time Change Data Capture (CDC) streaming into BigQuery with partitioned physical modeling and LookML semantic metric layers.

50k+ eps

Throughput

38%

Query Cost Reduction

99.9%

Data Quality Score

BigQuery CDCPySparkDataplexLookMLCloud StoragePyTest

MCP TOOLINGProduction

Cloud Run Microservices & Semantic Agent API

Containerized, auto-scaling API services on Google Cloud Run delivering low-latency semantic search and analytical agent endpoints.

< 800ms

Cold Start

99.99%

Availability

Thousands req/min

Throughput

FastAPIGoogle Cloud RunDockerCloud SQLPythonGCP

Work Experience

A decade of engineering distributed data systems, enterprise lakehouses, and production Agentic AI across Google, Carelon, Deloitte, and Cognizant.

AI & Data Engineer

Google

Hyderabad, India•2022 – Present

Current Role • Featured Impact

Partnering with major banks, financial institutions, and global enterprises to architect mission-critical Big Data lakehouses and deploy production Agentic AI systems with strict security guardrails.

Key Engineering Deliverables & Impact at Google

Engineered production agentic AI workflows and Model Context Protocol (MCP) database toolboxes for financial workloads with AST-level query validation.

Designed sub-second GraphRAG retrieval pipelines combining Cloud Spanner Graph and BigQuery for complex multi-hop financial knowledge reasoning.

Implemented high-volume CDC streaming and lakehouse physical models processing 50k+ events/sec benchmarked against enterprise TPC-DS workloads.

Pioneered end-to-end LLM agent observability using OpenTelemetry to capture reasoning traces, latency, token economics, and evaluation scores.

Core Stack:BigQueryCloud SpannerPySparkLangGraphModel Context ProtocolVertex AIOpenTelemetryCloud Run

Previous Enterprise Experience

Select a company to view role details

Senior Data Engineer

Carelon

Engineered scalable Big Data pipelines and cloud analytics infrastructure for massive healthcare and enterprise workloads.

2021 – 2022Hyderabad, India

Built distributed PySpark pipelines processing multi-TB transactional healthcare and claims records.

Engineered automated ETL workflows and validation frameworks ensuring zero data drift.

Optimized query runtimes and partitioning models on distributed storage for executive reporting.

Collaborated across international engineering teams on production cloud data platforms.

PySparkApache SparkPythonCloud PlatformsSQLAirflow

Articles & Writing

Deep dives published in Google Cloud on Medium covering Agentic AI, OpenTelemetry observability, and cloud data migrations.

Google Cloud8 min read

2025

Mastering Observability in Google ADK: Implementing OpenTelemetry for GCP Metrics, Traces, and Logs

A comprehensive guide to capturing distributed spans, metrics, and agent reasoning traces in Google Agent Development Kit (ADK) using OpenTelemetry and Cloud Trace.

Google ADKOpenTelemetryCloud TraceAgentic AI

Read on Medium

Google Cloud6 min read

2024

Hive to BigQuery: Move Data Efficiently Using GCP Dataproc Serverless

Architecting high-throughput, cost-effective data migrations from on-premise Apache Hive metastores to Google BigQuery using serverless PySpark on Dataproc.

Google BigQueryDataproc ServerlessApache HivePySpark

Read on Medium

Published on Medium