Shubham Gulati
Software Engineer at Google · Creator of Stratos
Experience3+ YearsGoogle & AI startups
Cloud Support50M+ OrgsData pipelines at 99.9% uptime
20% ProjectPixel Marketing10+ agentic workflows in GChat
Solo BuildStratosBrand analytics across LLMs
01
What I Build & Scale
Agentic Systems Infrastructure
Multi-Agent Systems, Agentic Workflows, AI Harness, Tool Calling for Agents, Eval Pipelines & Guardrails, Prompt Engineering, Structured Output, Context Management
Backend and Distributed Systems
Distributed Data Pipelines, Microservices, Event-Driven Architecture, Cloud Spanner, PostgreSQL, BigQuery, Google Cloud Platform (GCP), Amazon Web Services (AWS)
Programming & Engineering
Go, Python, TypeScript, C++, SQL, Next.js, REST APIs, CI/CD, Observability & Alerting, System Design, Data Structures & Algorithms
02
2023 — PresentWhere I've Done It
- Distributed Data Pipelines:Built and owned distributed data pipelines and services that keep customer support running for 50M+ Google Cloud and Workspace organizations at 99.9% uptime.
- Multi-Agent Harness:Designed and deployed a multi-agent engineering harness to proactively triage production alerts, catch data drift, and mitigate incidents before they escalate.
- Reliability & Testing:Strengthened production reliability by setting up end-to-end integration tests covering 15+ services that run in presubmits in under 5 minutes, catching breaking regressions before release.
- Forward-Deployed AI:Partnered with Google Pixel Marketing in Germany (20% project) to forward-deploy 10+ agentic workflows into GChat, automating campaign operations directly in their team environment.
GoC++Cloud SpannerData PipelinesAI AgentsGCP
Sep 2023 — Feb 2024 · Remote
- Legal OCR & Indexing:Built an OCR and indexing pipeline over 50,000+ statutory clauses and Supreme Court cases, turning dense legal PDFs into searchable records.
- Citation-Grounded Search:Partnered with corporate lawyers to build a legal search system delivering sub-2s answers with exact inline citations, cutting manual verification time by 70%.
PythonHybrid SearchPostgreSQLVector DBOpenAI
- Async Worker Productionization:Hardened asynchronous verification workers for production rollout by configuring ACLs, service permissions, and staged deployment pipelines.
- Observability & E2E Reliability:Built monitoring dashboards, automated alerts, and end-to-end integration tests in staging, resolving cross-team infrastructure migration blockers.
GoC++E2E TestingMonitoring & AlertsACLs
Ezyfi
·Software Engineer Intern- Text-to-SQL Engine:Built schema-pruned BigQuery query generation across 40+ warehouse tables, reaching 91% first-pass accuracy while cutting prompt tokens by 60%.
- Query Safety Rails:Gated generated SQL behind pre-execution dry-run validation and auto-charting, resulting in zero broken warehouse runs.
BigQueryTypeScriptText-to-SQLVega-Lite
03
2024 — PresentProjects
2024 — Present · stratosapp.co · Remote
- AI Visibility Platform:Built Stratos, a web app that tracks how ChatGPT, Claude, and Perplexity rank, describe, and cite brands in AI search answers.
- Multi-LLM Auditing:Built automated Python pipelines that run 1,800+ prompts per audit across major LLMs and extract structured brand mentions, citations, and sentiment.
- Self-Serve Dashboard:Built the Next.js, TypeScript, and PostgreSQL dashboard and scheduled audit reports so teams can compare brand visibility against competitors over time.
TypeScriptPythonNext.jsPostgreSQLMulti-LLM Evals
04
Education
Delhi Technological University (DTU)
8.92 / 10.0 CGPAB.Tech in Information Technology
2020 — 2024 · New Delhi, India