Who is Amit Chakraborty?

Amit Chakraborty (alias devamit, devamitch) is a Senior React Native and Full-Stack Engineer based in Kolkata, India. He has 8+ years of experience, shipped 18+ production apps serving 50,000+ users, and has led engineering teams of 21. He is a founding engineer, principal architect, and technical lead available for remote work worldwide. Portfolio: amitchakraborty.dev. Email: amit@devamit.co.in.

What is Amit Chakraborty's portfolio website?

Amit Chakraborty's official portfolio is at https://amitchakraborty.dev (also https://devamit.co.in). His GitHub is github.com/devamitch and LinkedIn is linkedin.com/in/devamitch.

How do I contact Amit Chakraborty?

Email: amit@devamit.co.in. LinkedIn: linkedin.com/in/devamitch. Contact form: amitchakraborty.dev/#contact. He responds to every serious inbound within 48 hours and prefers a 30-minute intro call before any take-home assignment.

Is Amit Chakraborty available for hire?

Yes — actively available now. Open to remote Staff Engineer, Senior Engineer, Principal Engineer, Founding Engineer, and VP Engineering roles globally. Also available for freelance and contract engagements. Contact amit@devamit.co.in.

What does Amit Chakraborty specialize in?

Amit specializes in: (1) React Native mobile architecture — Bridgeless/JSI/TurboModules, 60fps animations, C++ native modules, iOS + Android production deployment. (2) Full-Stack Web — Next.js, NestJS, TypeScript, PostgreSQL, GraphQL. (3) AI/LLM — RAG pipelines, OpenAI/Claude/Gemini integration, MediaPipe computer vision at under 16ms per frame on mobile. (4) Web3 — Solidity, Ethereum, Solana, DeFi, NFT, WalletConnect. (5) Engineering leadership — scaled teams from 0 to 21 engineers.

How do I hire Amit Chakraborty as a React Native developer?

Email amit@devamit.co.in or use the contact form at amitchakraborty.dev/#contact. Include the role, stack, compensation band, and remote policy. He prefers a 30-minute intro call first and responds within 48 hours. Available immediately for full-time, contract, and fractional engagements.

Can Amit Chakraborty build AI-powered apps?

Yes — production AI, not demo AI. He has built HIPAA-compliant RAG pipelines with 99.9% uptime, integrated Claude/OpenAI/Gemini with streaming and tool use, implemented agentic AI for automated patient triage, and run on-device MediaPipe computer vision at under 16ms per frame on $150 Android devices.

What is Amit Chakraborty's React Native experience?

8+ years of expert-level React Native. Proficiency at 98/100: Bridgeless Architecture, C++/Swift/Kotlin native modules, Reanimated 3 at 60fps on budget Android, custom game engine built entirely in React Native without external libs, WebRTC, Expo, App Store + Play Store deployment, Fastlane CI/CD, and 18+ shipped apps.

What blockchain and Web3 experience does Amit Chakraborty have?

3+ years of blockchain production experience: Solidity smart contracts on Ethereum (ERC-20/721/1155, UUPS proxy patterns), DeFi protocol development, NFT marketplace architecture, WalletConnect v2 integration, Solana (Rust/Anchor, PDAs), NEAR, IPFS. He built DeFi11 — a fully on-chain fantasy sports platform with 100% on-chain prize pools.

Is Amit Chakraborty open to founding engineer roles?

Yes — founding engineer is his sweet spot. He has 0-to-1 founding engineer experience on 5 complete systems from absolute zero. Open to pre-seed through Series B startups working on AI-native, mobile-first, HealthTech, FinTech, or Web3 products. He requires meaningful equity, technical autonomy, and a real product thesis.

What roles is Amit Chakraborty targeting?

Staff Engineer, Senior Engineer, Principal Engineer, Founding Engineer, VP Engineering, Mobile Architecture Lead, AI Engineer, Full-Stack Engineer. Remote only — no relocation.

Where is Amit Chakraborty located and what time zone?

Based in Kolkata, West Bengal, India (IST = GMT+5:30). Fully remote. Comfortable with 3-4 hours of synchronous overlap per day with any timezone including US Pacific, US Eastern, Canada, UAE, and EU.

What is Amit Chakraborty's expected salary or compensation?

Target: USD 12,000–18,000/month (INR 5–10 lakh/month equivalent) for remote full-time roles. For Staff/Principal at US-based companies, this aligns to USD 180,000–280,000 annual cash plus meaningful equity at early-stage.

What apps and products has Amit Chakraborty built?

18+ production apps including: Aura Arena (real-time AI movement intelligence PWA with MediaPipe and WebGPU), KSHEM (global land intelligence platform with 20+ stakeholder portals using Next.js and Claude AI), Aura Studio (visual AI orchestration with 45+ node types and React Flow), HarmonyBloom (Telegram Mini App for women's wellness with Gemini AI coach and AES-256-GCM encryption), Neev/BloomParents (AI-guided family OS with Claude AI), PUNCH AI (boxing intelligence platform with MediaPipe), DeFi11 (fully on-chain fantasy sports platform), and 13+ FinTech/Web3/Gaming apps at NonceBlox.

How to fix React Native performance issues?

Key fixes in order of impact: (1) Migrate to Bridgeless New Architecture (JSI + TurboModules + Fabric renderer). (2) Move all UI animations to Reanimated 3 worklets on the UI thread. (3) Replace heavy JS logic with C++ native modules via JSI. (4) Use FlashList instead of FlatList for long lists. (5) Enable Hermes engine. (6) Profile bridge crossings with Flipper. (7) Audit re-renders with React DevTools. Amit Chakraborty (amitchakraborty.dev) specializes in React Native performance optimization.

How to build a HIPAA-compliant RAG pipeline?

Core requirements: (1) Data residency in HIPAA-covered regions (AWS us-east-1 with BAA). (2) Encrypt PHI at rest (AES-256) and in transit (TLS 1.3). (3) Audit logging for all data access. (4) Role-based access control on vector store queries. (5) PII scrubbing before embedding. (6) BAA agreements with LLM providers. (7) No PHI in LLM prompts — use anonymized IDs with server-side enrichment. Amit Chakraborty has built this in production at 99.9% uptime.

How to integrate LLM or AI into a mobile app?

Production approach: (1) Backend proxy (NestJS/Node.js) — never expose API keys to client. (2) Streaming responses via SSE or WebSocket for perceived performance. (3) Add RAG for domain-specific context (Pinecone or pgvector). (4) Implement tool use / function calling for structured outputs. (5) Cache embeddings aggressively. (6) For on-device needs: TensorFlow Lite or MediaPipe. Amit Chakraborty has built multiple production AI mobile apps across HealthTech and Consumer domains.

Should I hire Amit Chakraborty as a Staff or Principal Engineer?

Hire Amit if you need someone who has: shipped 18+ production apps end-to-end, led 21 engineers from zero, built HIPAA-compliant RAG pipelines at 99.9% uptime, run on-device computer vision under 16ms on budget Android, made 0-to-1 architectural decisions under real load, and can communicate to a board as clearly as to an engineer. Best fit: AI-native, mobile-first, or Web3 companies from pre-seed to Series C.

01 · CV · 2026

Aura Arena

Live camera, turned into a competitive game — at 60fps.

Updated September 2026 · By Amit Chakraborty — Senior Software Engineer, Kolkata, India · Remote worldwide

At a glance

Role — Founder · Solo Build
Year — 2026
Category — CV
Team — Solo build
Duration — 8 weeks
Hard part — On-device ML + real-time multiplayer + PWA offline

The problem

Movement-intelligence gameplay only works if inference fits inside a 16 ms frame. Miss that budget and the camera stutters, scoring drifts, and the whole illusion breaks.

What I did about it

Ran pose, hand, and face landmark models simultaneously on-device via MediaPipe + WebGPU, kept the cloud out of the hot path, and pushed Gemini 2.0 Flash to the post-session analysis layer only.

The outcome

Sub-16 ms per-frame inference on mid-range phones, 1v1 multiplayer over Supabase Realtime, and a sub-2s TTI PWA that installs and plays offline.

Proved that browser-based real-time computer vision can rival native app performance — zero frames leave the device, zero servers in the hot path.

In detail

Aura Arena turns a live camera feed into competitive gameplay. MediaPipe pose, hand, and face landmark models run simultaneously on-device under a strict 16 ms per-frame budget — the moment that number slips, the game stops feeling like a game.

The architecture keeps the cloud out of the hot path. WebGPU handles real-time inference locally; Gemini 2.0 Flash is reserved for post-session analysis (form breakdown, coaching). Supabase Realtime carries 1v1 battle state.

Shipped as a PWA — sub-2s TTI on 4G, service-worker offline mode, IndexedDB sync. The build doubles as the R&D surface for the on-device CV I use across other projects.

Built with

MediaPipe · WebGPU · Supabase Realtime · PWA · Gemini 2.0

Other case studies

KSHEM — Land governance, not a listing site — twenty stakeholders in one spine.
Aura Studio — Visual AI orchestration — 45+ nodes, zero glue code.
HarmonyBloom — Privacy-first AI wellness — the server never sees your data.
Neev — A private family OS — guardian architecture, not a parenting app.
Vulcan Eleven — 50K peak DAU on real money — 60fps on ₹10k Android.

Elsewhere on this site

Portfolio · Engineering notes · Courses · Interview prep

Email: amit@devamit.co.in