APAnuj Patel
Open: AI / backend / full-stack

Applied AI + Backend Engineer

I build AI products that ship.

LLM apps / FastAPI / agents / full-stack MVPs

FoundersStartupsAI teams
Try Smriti LiveGet in touch

Ship MVPs

Wire APIs + LLMs

Think product

Proof4 projects
Voice AI agentVoice-first

VANI

Voice-first assistant built on Sarvam AI -> talk, ask, get answers

Sarvam AIVoiceLLMsPython
Details →DemoGitHub
Voice health companionVoice + memory

Smriti

Voice-first health memory + companion built for families

VoiceMemoryLLMsPython
Details →LiveGitHub
AI content engine<10s loop

Content Repurposing Engine

Docs / URLs / PDFs -> content drafts

PythonFastAPILLMsStreamlit
LiveGitHub
Crypto trend engineLive signals

TrendChain AI

YouTube / HN / CoinGecko / news -> viral ideas

PythonLLMsCoinGeckoHN
LiveGitHub

FAQ

What role is Anuj best fit for?

Applied AI engineer, backend engineer, or full-stack engineer on a product team that needs someone who can turn rough ideas into shipped workflows.

Is he only an AI engineer?

No. AI is the product layer he is leaning into, but the core strength is backend and full-stack shipping.

What kind of problems should I contact him for?

LLM apps, API-heavy products, MVPs, internal AI tools, RAG workflows, agent prototypes, and product demos that need to become real.

Can you build voice agents with LiveKit and Sarvam AI?

Yes. VANI is built on Sarvam AI for multilingual voice, and Smriti is a voice-first health companion. I wire LiveKit realtime pipelines — STT -> LLM -> TTS — into FastAPI backends that handle memory, RAG, and product logic.

How much does a voice AI agent cost and how fast can you ship it?

Prototype in 5-10 days, production hardening after. Pricing depends on scope: a single voice workflow starts small, with managed hosting and usage-based LLM/voice costs kept transparent. Ask for a 15-min scoping call.

Do you work with startups and businesses outside India?

Yes. Remote-first — I work with founders, startups and teams globally. Overlap hours, async updates, and shipping a live demo you can share with users, not slides.

What makes this portfolio different?

It shows the work, the operating style, and the public learning loop in one place instead of hiding behind generic resume language.