sahil.khatkar()
> initializing portfolio…
> loading modules [react · nestjs · aws · ai] ok
> mounting 9 projects ok
> establishing connection ready
INITIALIZING SEQUENCE0%
Open to opportunities · Pune, India

Sahil Khatkar

Full-Stack Developer @

I build production AI products end to end — cross-platform desktop apps with native audio and offline speech models, serverless NestJS backends on AWS, and React frontends. I like shipping real software and putting LLMs to work inside it.

0+
yr full-stack @ AIThinkers
0
projects shipped
0+
AIT-Scribe builds
0
Anthropic certs
SCROLL ↓
// the stack I build with
ReactTypeScriptNext.jsNode.jsNestJSAWSDockerPythonTailwindRedisPrismaMySQL
AnthropicOpenAIFlutterElectronViteReduxFastAPIFirebaseSocket.ioGraphQLSwiftExpress
// live preview

Featured work

AIT-Scribe
AIT-Scribe preview
flagship · shipped product

AIT-Scribe

Cross-platform desktop voice-to-text. Hold a key, speak, and your words are transcribed and pasted at the cursor in any app — fully offline via faster-whisper, with a floating widget over fullscreen apps, voice-activity detection and wake-word support. Native macOS audio capture; iterated across 100+ builds.

Native macOSCore Audiofaster-whisperVADWake-word
recoverycart.app
RecoveryCart preview
saas · side project

RecoveryCart

Recovers revenue from abandoned Shopify carts with AI-generated recovery emails — connect a store in minutes, run automated campaigns.

Next.jsTypeScriptPrismaShopify API
$ cat experience.log

Experience

Full-Stack Developer · AIThinkers

May 2025 — Present
  • Built and shipped AIT-Scribe, a cross-platform (macOS / Windows / Linux) desktop voice-to-text app — system-wide hold-to-dictate that transcribes speech and pastes at the cursor in any app, with a floating widget over fullscreen apps.
  • Implemented offline transcription via faster-whisper (no API key), voice-activity detection and wake-word support, with native macOS audio capture (Core Audio / HALInputUnit); iterated across 100+ released builds.
  • Engineered the AIHR enterprise portal backend with NestJS, TypeORM and MySQL deployed serverless on AWS Lambda — S3, SES, KMS, Secrets Manager, real-time WebSockets (Socket.io), and Swagger-documented REST APIs.
  • Built the portal frontend in React + TypeScript + Vite with Redux Toolkit, shadcn/ui, TanStack Table and React Hook Form; documented components in Storybook with a11y checks and Vitest coverage.
  • Built xThinker, an AI agent-platform dashboard with an in-browser Monaco editor and a live execution sandbox via WebContainers, Auth0 auth and Recharts analytics.
$ ls ./projects

Projects

enterprise · full-stack

AIHR Portal

HR platform — a serverless NestJS API on AWS Lambda (TypeORM/MySQL, S3/SES/KMS, Socket.io, Swagger) with a React + Redux + shadcn dashboard and real-time updates.

NestJSAWS LambdaTypeORMMySQLReactSocket.io
AI agent platform

xThinker

Agent-platform dashboard with an in-browser Monaco code editor and a live execution sandbox via WebContainers, plus Auth0 auth and Recharts analytics.

ReactWebContainersMonacoTanStack QueryAuth0
fintech · data

Mutual Fund Analyzer

Indian mutual fund & market-intelligence system with a “smart-money” signal engine — ingests fund and market data to surface portfolio insights and analytics.

PythonMarket Data APIsAnalytics
platform · python

FileForge

AI-powered PDF intelligence platform — convert, extract and transform PDFs at scale with an async task queue.

ReflexFastAPICelery/RedisDocker
saas · side project

RecoveryCart

Recovers revenue from abandoned Shopify carts with AI-generated recovery emails — connect a store in minutes, run automated campaigns.

Next.jsTypeScriptPrismaShopify API
desktop · ai

AI Work Assistant

Electron desktop assistant pairing the Anthropic API with local Whisper transcription for voice-driven, context-aware help.

ElectronAnthropic APIWhisperPrisma
automation · python

Google Meet AI Agent

Joins a Google Meet as an authenticated user, gives a TTS intro, transcribes the call with Whisper, and answers questions via OpenAI.

PythonSeleniumWhisperOpenAI
generative · python

AI Video Pipeline

Turns a script into a narrated video: copy from the Anthropic API, voice from ElevenLabs, alignment with Whisper, assembly with MoviePy.

Anthropic APIElevenLabsWhisperMoviePy
$ ./load_arsenal

Arsenal

// languages

TypeScriptJavaScriptPythonDartSQL

// frontend

ReactNext.jsSolidJSViteRedux ToolkitTailwindshadcn/uiTanStack

// backend

NestJSNode.jsExpressFastAPITypeORMWebSocketsREST

// cloud & data

AWS LambdaS3 / SES / KMSServerlessDockerMySQLPrismaRedisFirebase

// ai / llm

Anthropic ClaudeMCPOpenAIWhisperElevenLabsClaude Code

// cross-platform

FlutterElectronNative macOS (Swift)
$ whoami

Background

// certifications · Anthropic

Claude Code in ActionAnthropic
Claude with the Anthropic APIAnthropic
Introduction to Model Context ProtocolAnthropic
Introduction to Agent SkillsAnthropic

// education

PG Diploma, Internet of ThingsC-DAC, Pune · 2024
B.Tech, Electronics & Communication EngineeringKurukshetra University · 2020–2023
Diploma, Electronics & CommunicationChotu Ram Polytechnic, Rohtak · 2018–2020