Karthik Vinayan

applied ai engineer

Building AI at Clueso (YC W23). Previously built the backend for a production AI cloud automation platform: multi-agent orchestrator, knowledge graph infra, MCP, semantic memory, all from zero.

Applied AI Engineer
Clueso, Bengaluru
Founding AI Engineer
AI & Backend Systems
Omni RPA Inc, San JoseAgentic Solutions Pvt Ltd, Hyderabad
Research Intern
Computer Vision
Digital University of Kerala, Kerala
podspawnDocsGitHubGoAGPL-3.0

One command gets you a dev environment, locally or over SSH.

Single Go binary on Docker. Composable Podfile config, branch-isolated workspaces, native sshd, gVisor sandboxing, and a session control plane with actor-scoped audit.

ergoGitHubTypeScriptBun

Code review from the AI subscription you already pay for.

Runs your linters locally and hands the findings to ChatGPT or Codex. The verdict comes back as TUI, JSON, SARIF, or markdown, with the token bill for every review.

dconGitHubGoHomebrew

The docker CLI, rebuilt on Apple's container runtime.

Every container gets its own lightweight VM. Warm-pool pre-boot cuts start time from ~700ms to ~90ms. One ~6MB static binary, on Homebrew.

hnGitHubNext.jsTypeScriptMIT

Hacker News, as the client I actually wanted.

Next.js 15 with RSC. Writes proxied through news.ycombinator.com with per-request CSRF scraping and rate limiting, encrypted iron-session cookies, IndexedDB highlights that survive edits via fuzzy anchors, collapsible threads, reply inbox, reader mode.

Languages: Python, Go, Rust, TypeScript, SQL, Java
Backend: FastAPI, WebSockets, NATS JetStream, Event-Driven Architecture
Databases: FalkorDB, Neo4j, Weaviate, Supabase (Postgres), Redis
Cloud & DevOps: AWS (CCP), Azure (AZ-104), GCP, Docker, Kubernetes, OpenTelemetry
AI & ML: Multi-Agent Orchestration, MCP Protocol, RAG, Semantic Caching, LLM Tool Calling, Eval Frameworks, pgvector, vLLM, llama.cpp

Building something with agents or infra? I read everything sent my way, usually same day.

Karthik Vinayan | Applied AI Engineer at Clueso