Theeraj Chandra
I’m a Computer Engineering student at the University of Waterloo, building backend systems and AI tools. Currently working on Alexa+ at Amazon.
Work
AmazonSep – Dec 2026 Software Engineer Intern · Alexa+ Inference evaluation infrastructure and autonomous prompt optimization for Alexa+.
- Built a serverless Java/AWS platform that cut failure-to-fix time from 2 days to 2 hours across 3 Alexa teams.
- Built an MCP agent that resolved 80% of critical failures within 3 iterations and reduced inference spend per round by 85%.
- Moved inference to an asynchronous Lambda/SQS pipeline, reducing a 500-inference run from 40 minutes to 4 minutes. Added versioned DynamoDB/S3 persistence, leases, and snapshots.
StackAdaptJan – Apr 2026 Software Engineer Intern · QE Infrastructure AI agents, developer tooling, and infrastructure for engineering teams.
- Built a TypeScript PR-review agent with Bedrock, MCP tool calling, structured output validation, caching, and OIDC authentication.
- Diagnosed a PostgreSQL bottleneck on a 45M-row table and drove a partial-index optimization, reducing lookup latency from 25 seconds to 0.919 milliseconds.
- Built a feedback loop that used PR outcomes and reviewer feedback to guide future agent fixes. Consolidated GraphQL logic across 8 integration specs, removing 1,240 duplicated lines.
SuperWorldJan – Apr 2025 Software Engineer Intern Conversational maps, location-aware experiences, and product analytics.
- Built UI flows for a LLaMA 3-powered conversational map, enabling location discovery and navigation through natural language.
- Shipped geolocation check-in with Haversine distance calculations and a 300-metre proximity check.
- Built a React/TypeScript analytics dashboard with parallel API calls and live visualizations for the founders.