Neeraj · Full-Stack AI Engineer & Voice AI Developer (Delhi, India)

What I do

Three lanes. Same standard of work.

Production voice systems, agentic backends, and the occasional 1:1 session. Pick the lane, and I'll meet you there.

Voice AI engineering

Production voice agents at 1.2s full-turn latency. Real-time orchestration, telephony, barge-in, multi-tenant isolation, and observability, built for systems that actually answer the phone.

1.2s full-turn latency
GDPR-grade isolation
Multi-tenant under load

LiveKit
Azure AI Foundry
Realtime
Telephony

From spec to production

POCs first. Scoping with numbers. Production runtime. Evals before traffic. Team handoff.

POCs first. Prove the approach end to end before committing to the full build.
Scoping: latency budget, success metrics, and constraints written down.
Production runtime: LiveKit, LLM, STT and TTS, telephony, tool use, all wired.
Eval harness and observability shipped before any production traffic.
Handoff with docs and hands-on team enablement so your engineers own it.

Hire me for voice work

AI agents & full-stack

Tool-using agents and RAG pipelines shipped as real products. Eval harnesses, hybrid retrieval, and the full-stack to ship them. TypeScript and Next.js up top, Node.js or Express on the API, Python for the AI/RAG layer.

Agents
RAG
TypeScript
Next.js
Node.js
Express
Python

Learn more

Mentorship & sessions

1:1 sessions for AI builders, hackathon mentoring, college talks, and workshops. Honest, focused, no generic advice.

1:1 sessions
Hackathons
Workshops
Talks

See sessions

Track record

Where I've been shipping.

IntellifyAI

Full Stack Engineer (Contract)

Sep 2025 to May 2026

London, UK · Remote

Shipped an enterprise voice AI platform to production, hitting ~1.2s median full-turn latency (STT, LLM, TTS, VAD combined) with sub-second per-stage targets.
Load tested the platform at ~10 concurrent tenants with isolation verified, meeting the multi-tenancy and GDPR bar required for enterprise sign-off.
Designed and built the platform RBAC system end to end on my own in TypeScript on Node.js and Express, covering role and permission modelling, enforcement across services, and the admin surface to manage it.
Architected the backend from zero with the founder: Node.js and Express in TypeScript for the main platform (business logic, RBAC, admin, tenant management), Python with FastAPI for the LiveKit voice agent runtime and deployment pipelines.
Validated the approach through multiple POCs before committing to the full build, then automated per-tenant Azure rollout in Python so onboarding became a repeatable pipeline rather than a manual ops job.
Led a small team of senior engineers through delivery, then guided them after launch so they could own the platform once I rolled off.

TypeScript
Node.js
Express
Next.js
LiveKit
Azure AI Foundry
Python
FastAPI
Docker
+5 more

Mindcraft Labs

Fullstack AI Developer

Feb 2025 to Sep 2025

Gurgaon, India · Remote

Converted multiple multi-step manual onboardings into chatbot and voice agent flows, collapsing human-handled intakes into self-serve conversations.
Engineered three voice agent platforms on top of whitelabel providers (ElevenLabs, Retell AI), extending each in TypeScript on Node.js with custom logic, integrations, and workflow scaffolding so they fit real business processes rather than running as stock agents.
Built a Sheets-to-call automation that placed bulk outbound calls, captured structured results, and triggered downstream workflows, handling ~100 calls per campaign.
Shipped Node.js microservices, Python scripts, and low-code workflows across the wider automation portfolio to remove daily ops time across the business.
Owned full-stack delivery on Next.js, Node.js, and Firebase in TypeScript, from data model to UI to monitoring.

TypeScript
Node.js
Next.js
Firebase
Python
ElevenLabs
Retell AI
+3 more

Menrva Technologies

Front-end Developer Intern

Oct 2023 to Jan 2025

Bangalore, India · Remote

Delivered agile React development for product interfaces, integrating AI services via APIs with full user-management flows.
Owned UI feature delivery from Figma to production, including responsive layouts and accessibility.
Collaborated directly with the founder and leadership on front-end strategy, contributing to project delivery and product decisions.
Recognised as Best Intern of the Quarter (Oct to Dec 2023) for shipped feature work.

React
TypeScript
CSS
Firebase

Proof of work

Selected projects.

A few of the systems I've shipped. Detailed write-ups are coming as soon as they clear NDA. In the meantime each card links to the long-form study.

Talk to my portfolio

Live

on this site

A voice agent that drives this site over RPC: ask it to open a case study, book a call, or download the resume, and watch the page respond.

LiveKit
Python
Next.js
RPC
Voice AI
+1 more

Read case study

Enterprise voice AI platform

1.2s

median full-turn latency

Shipped a multi-tenant voice agent platform to enterprise production, with 1.2s median full-turn latency and GDPR-grade isolation across tenants.

TypeScript
Node.js
Express
Next.js
LiveKit
Azure AI Foundry
Python
+3 more

Read case study

GoReach

2.3x

lift with voice setup

A LinkedIn outreach product that writes in the user's own voice via per-user RAG. Live since April 2026 with 102 beta users and a 2.3x quality gap between voice-setup-complete and skipped.

TypeScript
Next.js
Node.js
Python
FastAPI
RAG
AWS
+3 more

Read case study

Sheets-to-call automation pipeline

1 support role

absorbed into automation

A Google Sheet automation that turns rows into parallel voice calls and writes structured outcomes back. Built at Mindcraft Labs to absorb a support team's manual dialling work without making them leave their workbook.

Python
Google Sheets API
Webhooks
ElevenLabs
Workflow Automation
+1 more

Read case study

Live experience

Talk to my portfolio.

A real voice agent walks you through this site, opens case studies on your command, books slots over a call, and answers questions about my work in real time. Click below to start.

Voice tour

Live

LiveKit agent with sub-second latency. Function tools wired into the site.LiveKit realtime agent with sub-second per-stage latency, an English turn detector, and function tools wired into the site. Same stack featured in the enterprise voice AI case study.

Navigates the site
Opens case studies
Books a call
Downloads the resume
Answers questions live

or press/from anywhere

Stack

Tools I reach for.

TypeScript on Node and Next.js for product, Python for AI, RAG, and voice. Hover any badge for a quick note on how I've used it.

Languages

TypeScript
JavaScript
Python
SQL
C / C++
HTML
CSS

AI / ML

LLMs
RAG
Agentic AI
Function Calling
Voice Orchestration
LiveKit
Azure AI Foundry
Deepgram
Cartesia
ElevenLabs
Retell AI
Embeddings
Prompt Engineering

Frameworks

Next.js
Node.js
Express
React
FastAPI
Tailwind CSS
Framer Motion

Cloud / Infra

Azure
AWS
Vercel
LiveKit Cloud
Docker
GitHub Actions
Linux

Data

PostgreSQL
pgvector
MongoDB
Convex DB
Firebase

For students & devs

Want a second pair of eyes on your AI work?

Free 30-minute calls. Bring a real problem and we will dig in.

I ship voice agents enterprises pay for. That means I can tell you which parts of the AI stack actually matter in production, and which tutorials are wasting your weekends.

What we can dig into

AI career roadmap: what to build, what to skip, what hiring managers actually look for.
Voice agent architecture: latency budgets, eval harnesses, observability, multi-tenant scale.
Portfolio and resume teardown: concrete edits, line by line.
Whatever you're stuck on. Bring the actual problem.

Free

Not paid. Just a call. If you want deeper ongoing work after, we can talk about it on the call.

For event organisers

Talks from an engineer who actually ships voice AI.

Hackathons, bootcamps, college lectures, industry panels. Based in Delhi, India. Happy to travel across India and worldwide, or join remotely.

Neeraj on stage facing an audience during a speaking session — On stage with the audience in view.

Most AI talks are slides about demos. Mine come from a production voice platform: ~1.2s median full-turn latency, multi-tenant isolation under load, GDPR-grade. Audiences leave with what the textbooks miss.

War stories from shipping enterprise voice agents, not recycled tutorials.
Concrete numbers: latency budgets, eval harnesses, tenant isolation tradeoffs.
A working engineer who can answer the hard questions in the Q&A.

See formats, sample topics, abstracts, and past events

Got an event in mind?

Share format, audience, and date. I'll get back within a few days.

For founders & companies

Production systems that ship to enterprise, not demos.

Full-stack engineer for product work, voice AI, and agentic backends. Sub-second stage latency, GDPR-grade tenant isolation, and the engineering to back it all up.

Just shipped an enterprise voice AI platform: ~1.2s median full-turn latency, ~10 tenants verified isolated under load, GDPR-grade and security-reviewed. Available now for new work.

Just want the basics?

Skip the calendar. Email me, grab the resume, or reach me on LinkedIn.

Free, no commitment. We'll figure out scope and fit on the call.

How I work with teams

Voice AI from spec to production

Take a voice agent from whiteboard to enterprise-ready. LiveKit, Azure AI Foundry, Node.js, FastAPI, Next.js. Sub-second stage latency, tenant isolation, eval harness, observability, handoff docs. Delivered, not demoed.

Fractional AI engineering lead

Embedded with your founding team. Architecture calls, code review, hands-on work on the critical path, and team mentoring. For startups that need a senior voice or AI engineer without the full-time hire yet.

Honest discovery call

Not sure what you need? Book the call. I'll give you a straight read on what should be built, what shouldn't, and whether I'm the right person to build it.

Get in touch

Tell me what you're working on.

Pick a reason and I'll get the right context up front. Replies usually within a few days.

Just want to say hi? DMs open on LinkedIn and X.