Lohith Kumar

By day, enterprise software. By night, my own products, with AI in the loop.

Full-stack engineer. Java, TypeScript, Python. Three years in enterprise systems, three live products, AI in production rather than in slides. Open to full-time roles, remote or anywhere in the US.

Day

  • 2026 → nowMastronardi Produce · Full-stack engineer. Grower Pay, the grower accounting and settlement platform.
  • 2025 → 26Marvel Technologies Inc. · Java, then AI engineering. Demand forecasting carried through to billing.
  • 2024 → 25Anywhere Real Estate · Spring Boot on Amazon Bedrock for listings. Throughput up 30%.
  • 2024Greenlots (Shell) · Notification services. Intern.
  • 2022Infosys · Device trust and risk scoring.

5:30 pm. Laptop closes. Laptop opens.

Automations

  • Listing copy pipeline: property photos in, tags, captions and descriptions out. Amazon Bedrock, in production at Anywhere Real Estate.
  • Demand-to-billing pipeline: Prophet forecasts feeding purchase orders, inventory and journal entries. Built at Marvel Technologies.
  • Contract review pipeline: an LLM reads the clauses, deterministic rules decide. Manual review time down 90%.
  • Phone booking agent: answers the call and takes the caller through a booking. Retell, n8n.
  • Automation suite: 36 n8n workflows, 586 nodes, one Postgres schema. Reminders with cancellation backfill, receivables chase, speed-to-lead, review replies with a human approve step, invoice OCR.
  • Ad production pipeline: a product reference in, faceless spec reels out. Seedance, driven by a set of Claude skills I wrote.
  • This site: drafted from my Obsidian vault by Claude as an operating partner, fact-checked line by line, shipped static.

How I build these