Shrithan Devaiah

Bengaluru, India (Open to Remote) | +91 63628 21846 | shrithanofficial@gmail.com

Technical Skills

Languages & Front End | TypeScript, JavaScript, Python, React, Next.js, React Native, Tailwind CSS

Back End | Node.js, FastAPI, PostgreSQL, REST, GraphQL, WebSockets, Express

AI & Agents | Mastra, CrewAI, LangChain, MCP, RAG, OpenRouter, Inngest, Temporal

Cloud & Infra | AWS (ECS, Lambda, SQS, DynamoDB, Bedrock), Docker, CI/CD, Vercel, Railway

AI-native email and CRM client. Seed-stage, ~10 engineers.

Architected the core agentic system from the ground up, a Mastra multi-agent runtime serving 250,000+ agent tool calls a month, with model routing and failover across four providers
Took the agent beyond web chat into six channels (Slack, iMessage, Telegram, email, LinkedIn, X), owning the routing and auth layer that lets one agent hold a conversation on any of them
Built the agent's memory layer, observational recall, a user-scoped working store, and durable facts across threads, 95% on LongMemEval
Built the agent's cloud coding sandbox end to end, where a coding agent writes, runs, and previews full-stack apps and publishes them as live sites, with long builds on durable jobs that outlive the request
Extended Prism, our typed object graph over Postgres, so the agent can traverse CRM relationships, and built the HubSpot and Attio import that avoids duplicates

Cut a two-week Ahrefs sprint to same-day keyword and competitor pulls, as tools the CrewAI crews run for agencies
Ran those crews on a FastAPI server with Temporal, so chat and SEO jobs resume after failure
Gave the crews clustering, content briefs, and backlink gap analysis, with a human approving before ship
Shipped the React app agencies use, keyword research, a Kanban of agent tasks, and a chat copilot

AI education platformReact | Node.js | GraphQL | PostgreSQL

Shipped the core education platform in 14 days, GraphQL over Postgres with progress tracking and analytics
Modeled courses, users, and quizzes on one GraphQL schema, with feedback and gamification in the same product
Added AI mock interviews next to lessons, so practice lived in the same product as progress

Cut release time from 4.2 days to under an hour, with CI/CD
Served 40k+ food-delivery requests a day, a Node.js API for order create, read, and status
Pushed order and delivery status over WebSockets, so the client updated as the order moved

Built a multi-tenant SEO agent in Slack, approval-gated audits and 30-day plans, tenant-scoped data and encrypted OAuth

VTU - Bachelor of Technology, Computer Science