Nagarjun Mallesh.

Professional outreach is usually slow, generic, or unsafe because it ignores the real context users already have in profiles, resumes, and intent.

A shipped full-stack AI product with a Chrome MV3 extension, authenticated dashboard, resume/profile management, and reviewed LinkedIn/email draft workflows.

Service workers, content scripts, side-panel UI, Supabase Auth/RLS, credit enforcement, rate limits, structured Claude outputs, prompt-injection defenses, Sentry, Vitest, and Playwright.

Shows the founder-engineer loop: identify a painful workflow, design the product surface, build the full stack, add safety rails, and ship it for real users.

Resume tools often give generic advice or fabricate improvements candidates cannot defend in interviews.

A FastAPI and Next.js platform for parsing resumes, recruiter-style evaluation, JD tailoring, versioning, preview, and country-aware PDF export.

Schema-constrained LLM responses, hallucination guards, ATS simulation, Supabase auth/storage, credit accounting, and Playwright PDF rendering.

Turns an ambiguous career-support workflow into a defensible product system with clear user value, evaluation boundaries, and production delivery.

System design topics are often explained with static diagrams that hide runtime behavior, failure modes, and the tradeoffs engineers actually debug.

A client-side simulator with interactive modules across databases, Kafka, TCP, rate limiting, transformer inference, sharding, and scaling patterns.

Typed simulation engines, state-driven controls, SVG visualizations, live metrics, and module-level architecture for adding new system internals.

Turns abstract infrastructure concepts into inspectable systems for engineers who learn better by watching behavior unfold.