Vol. INo. 1May MMXXVI

New York · Brooklyn · WorldwideSingle Copy · Free

Rohan Chawla

The Builder's Desk

¶ The Lead

Software engineer building Sanji, a localhost replacement for NotebookLM. Joining Intuit as SWE in June 2026 after a Software Infrastructure internship at Verrus, an Alphabet-backed AI infrastructure company. Previously at John Deere (twice) and NOAA. Most interested in agentic RAG, retrieval, and the path to AGI.

Selected Work
§

Sanji

A study buddy that lives on your laptop

Sanji is a pet you study with. Drop a PDF, a DOCX, or a folder of markdown notes in front of him and he'll read it all, sit with you while you work, and answer anything you ask about it. Your notes never leave your machine. Your study buddy never leaves your desk.

Agentic Brain

A second brain

Notes, calendar, projects, and decisions in one markdown vault, with rituals that actually run. Morning planning that reads my calendar and packs the day. Weekly and monthly rollups over tagged daily captures. A drift skill that compares stated priorities against what my git history and calendar actually show me doing. Kept three concurrent tracks alive through senior year.

Experience
§
2025 — 2026

Verrus

Software Engineering Intern
Alphabet-backed AI infrastructure

Built the company's first automated continuous deployment pipeline (ArgoCD, Helm, Harbor) for the policy server on on-prem Kubernetes. Stood up monitoring across a 6-node distributed NATS cluster with NATS Surveyor and Prometheus. Shipped 6 core UI features on the marketing site that unblocked the product team from engineering hand-offs.

Summer 2025

John Deere

Software Engineering Intern
Dealer-notification pipeline

Led the migration of the Spark jobs that alert dealers to damaged customer machines onto Microsoft Fabric, cutting job run costs 73% and saving $676K per year. Built Chalice middleware around MultiDict with 100% unit-test coverage, hardening the API path behind it.

Summer 2024

John Deere

Software Engineering Intern
ML model serving

Rebuilt the ML model deployment path on Databricks Model Serving, AWS Lambda, and Terraform. Cut release cycle by 2+ days, cut deployment failures 30%, sustained 16,500 requests per second. Tuned Databricks model-serving features for $14K in hosting savings.

Summer 2023

NOAA

Software Engineering Intern
Arctic voyage planning

Modeled 14.7M Arctic grid cells and computed least-cost sea-ice routes via Dijkstra (NetworkX, NumPy, NetCDF4), cutting simulated transit time 20% on Northwest Passage scenarios. My routing module was integrated into NOAA's global voyage-planning system for worldwide pathfinding.

Coming Up
§
June 2026

Intuit

Software Engineer 1
QuickBooks Indirect Tax, New York

Joining QuickBooks Indirect Tax in NYC. Building AI features inside the tax-compliance platform: anomaly detection, discrepancy analysis, and filing recommendations.

Correspondence
§
GitHub
github.com/rohanc2k4 →
LinkedIn
linkedin.com/in/rohanchawla2004 →
Email
rohanchawla2004@gmail.com →