Hi, I'm Shreekant Mandvikar.I (actually) build GenAI & Agentic AI solutions.

A bedtime-story generator built for actual family use, not a demo: give it a child's name, age, and interests, and it writes and illustrates a personalized story in seconds — the same engineering rigor I put into enterprise agents, aimed at a much smaller, much more honest audience.

It's the project that keeps the rest of my work grounded — every retrieval pattern, guardrail, and prompt idea gets stress-tested here first, on an audience of one very direct four-year-old critic, before it goes anywhere near production scale.

A six-part series that takes builders from first principles — what an agent actually is, why it's different from a chatbot — through to production readiness: setup, tools and memory, deployment, specialized agents, and scaling.

Taught through LinkedIn, where it reached the same audience that engages with my weekly writing on agents, RAG, and LLM systems — all six parts are archived, with real numbers, on the writing page.

An agent that lives inside a message thread, not an app: text it and it reads your LinkedIn and Instagram activity, drafts things like a birthday post before you forget, and keeps a running plan for your life instead of a to-do list you never open.

It's the same agent-lifecycle pattern behind everything else I build — identity, scoped access to your accounts, a plan it updates rather than a script it replays — just pointed at a life instead of an enterprise. Full write-up on the writing page.

A cross-platform assistant that's always one gesture away: double right-click anywhere and it captures a screenshot and 15–20 seconds of audio, transcribes and encodes both, and forwards them to an LLM — Groq by default, OpenAI as a fallback — for a context-aware answer to whatever you're looking at.

No app-switching, no copy-pasting your problem into a chat window — just the trigger, the context it already has, and an answer. Built with Tkinter, `pynput` for the global click listener, and `sounddevice`/`SpeechRecognition` for audio.