Bhavish.ai
Autonomous Digital Clone Console
Bhavish.ai
Hey, I'm Bhavish 👋
Full-Stack AI Developer
Welcome to my interactive console! You're chatting directly with my digital clone. Ask me about my architecture choices, check out my work history, or demo my live projects below!
Full-Stack AI Developer
Full-TimeProductHunt #3 Product of the Day- Engineered company VS Code Extension for mobile app testing with custom token storage & auto-refresh mechanisms.
- Built real-time low-latency WebSocket & ADB web UI stream for remote Android & iOS devices with interactive touch gesture relay (BrowserStack competitor).
- Built custom on-device gRPC server & SSE backend pipeline, extracting Android UI tree data to render a custom interactive Mobile UI Inspector for Maestro code generation.
- Integrated Appium Inspector via proxy and added vision AI agents generating Appium test scripts automatically.
- Designed Self-Healing test runner capturing findElements calls and recovering failed locators using visual similarity matching.
- Managed version control with Git/GitHub and used GitHub Actions to automate build, test, and deployment (CI/CD) for production releases.
Full-Stack AI Developer
Full-TimeProductHunt #1 Product of the DayFrontend Developer Intern
InternshipJobPilot - AI Job-Search & Application-Prep Platform
Resumable LangGraph multi-agent pipeline that drafts tailored resumes, cover letters, and interview prep grounded in the user's real background — deliberately prepare-only, it never auto-submits on the user's behalf. Distributed, atomic Redis rate-limiting (Lua-script token bucket + daily-quota gate) protects a shared LLM budget across a Celery worker pool, and Postgres-checkpointed state resumes failed runs from the exact step that failed, never from scratch.
AskDocs - RAG Document Q&A Platform
Self-correcting RAG chat pipeline in LangGraph over a pgvector vector database with Gemini embeddings (retrieve → grade → conditional rewrite → generate → verify) that grades its own retrieval quality and checks answers for groundedness before returning them. Crash-safe async ingestion via a hand-written Postgres job queue (FOR UPDATE SKIP LOCKED), from-scratch auth with per-user data isolation, and a token-by-token SSE streaming UI in vanilla JS — backed by 64 automated tests running in GitHub Actions CI against a real Postgres instance.
Resonance - Full-Stack AI Voice Generation SaaS
Multi-tenant AI voice generation platform powered by open-source Chatterbox model on Modal serverless GPUs. Features Paddle billing, S3 audio storage, workspace management, and custom session/OAuth auth.
Autonomous Vision-Agentic UI Test Pipeline
Autonomous vision-agentic testing pipeline combining multimodal LLM perception with visual grounding. Executes multi-step browser actions from natural language intent while live-streaming auto-generated Playwright test scripts.
NativeBridge Remote Device Farm & Mobile Inspector
Real-time remote mobile device farm (BrowserStack competitor). WebSockets + ADB touch gesture relay for manual device control, plus custom on-device gRPC server & SSE pipeline rendering an interactive UI node inspector for Maestro code generation.
Self-Healing Tests
Automated testing proxy capturing Appium findElements calls. Records passing locator snapshots on initial runs and uses visual similarity matching to automatically recover broken locators on test failures.
Frontend & Desktop Engineering
Backend & Streaming Systems
Backend Infra & Messaging
AI Agents & Testing Automation
DevOps, Cloud & Storage
M.Sc. Computer Science
GraduatedAdvanced computer science curriculum focused on distributed systems, algorithm optimization, software engineering patterns, and artificial intelligence.
I'm actively seeking Full-Stack & AI Systems Engineering roles! Whether you want to discuss my vision-agentic testing pipeline, BrowserStack competitor device farm, self-correcting RAG pipelines, or full-stack SaaS architectures—let's connect.