Incoming internship on the Microsoft AI team, focused on Copilot Tasks and Growth.

• Engineered and deployed a LangGraph-based AI log triaging agent for infrastructure diagnostics to analyze Systemd and Kubernetes logs, reducing manual RCA time from days to minutes.
• Implemented self-correcting workflow orchestration with sub-agent spawning, parallel execution, and context preservation, enabling automatic failure recovery and workload distribution across specialized agents.
• Developed a lightweight log streaming pipeline to Elasticsearch using FluentBit to enable structured log querying for AI agents, deployed across thousands of clusters, achieving sub 20MB memory usage.
• Integrated MCP configuration options enabling AI agent routing to Claude Code CLI with local MCP servers (Fuse, Elasticsearch, Kubernetes, Sequential Thinking), providing users enhanced tooling capabilities.

• Developed an AI SaaS to allow users to create custom enterprise software with no code during the W24 YCombinator batch, shipping 570+ features for 200+ paid users, contributing to successful Series A funding.
• Maintained a native AI chatbot to suggest workflows using LangChain with function calling and structured JSON outputs, reducing latency from 24 to 7 seconds via prompt caching and dynamic LLM fallbacks.
• Architected a trigger service to enable automatic workflow execution on Google Kubernetes Engine, leveraging GCP Cloud Functions with CRON jobs for refreshing webhooks and Redis for in-memory caching.
• Parallelized workflows using Celery and Redis for worker management, enabling multiple concurrent workflow steps per user and optimizing runtime by up to 1500%.
• Optimized Google Spanner SQL queries by indexing and batch processing, reducing latency by over 1900%.
• Designed and developed a responsive landing page using React and Redux Toolkit for state management and TailwindCSS for styling, increasing conversion rates by over 1700%.
• Integrated with 30+ providers including OpenAI, Slack, Notion, Airtable, and Salesforce using Python and OAuth2.0, allowing users to link workspaces, CMS, and AI services to build comprehensive workflows.