Skip to main content
Ajay Prajapat / engineer, architect, builder

Complex ideas.
Working systems.
Built with clarity.

I connect interfaces, APIs and AI into products with a solid foundation. From the first design to the systems behind it.

Ex-Apple Sr. Fullstack Engineer15+ years · Platform & agentic engineering
Scroll to see how I think, build and ship
Ajay Prajapat seated with a laptop, surrounded by code and system connections
Sr. Fullstack Engineer at Apple, 2022–202615+ years shipping production softwarePlatform + agentic engineeringAngular 9–10y · Node.js 9–10y · TypeScript 11–12yFounder of APPNEURAL since 2021Live sessions every Sunday on Innovate with Ajay
Ajay seated in a chair, holding a laptop on his lap
ajay.os — clinic-bookings
CLINIC / TODAYBookings, sorted.✓ Patient confirmed · 11:30✓ Reminder scheduled● System online
@Tool('findSlots') async findSlots(day, doctor) {   return this.calendar.free(day); } @Tool('book') async book(slot) {
Ex-Apple · 4 yrs Platform + Agentic engineering
Inside the screen · from idea to live

Ajay is building
something.

One laptop. Every layer.
Scroll inside and watch an idea become a working product.

Scroll to enter the workspace
ajay.os — example project: clinic bookings
01 / THE BRIEF

Start with the real problem.

A clinic needs fewer booking calls. We agree on what the product must solve.

01 · BriefI start by listening. You explain the problem in your words, and I write down what done looks like.
● REC00:00Discovery call · Clinic owner × Ajay Hindi / EnglishNotes on
CO
Clinic owner
Ajay on the callAjay● NOTES
WHITEBOARD · PROBLEM"Front desk spends the day answering booking calls."
The conversation

ClientBooking calls take all day. Missed visits leave empty slots.

AjayWhere do patients usually reach you?

ClientWhatsApp. Hindi + English. No reminders.

AjayLet's book and confirm in chat. Staff can take over.✓ Booking · ✓ Reminders · ✓ Staff control
Scope agreed ✓

Requirements · v1
Goal
Patients book and confirm on chat, without calling.
Must-haves
  • ✓Booking on chat, any hour
  • ✓Reminder the day before
  • ✓Staff can take over any chat
  • ✓Today's schedule on one screen
Constraints
WhatsApp firstHindi + EnglishNo new app to install
Done looks like

Fewer booking calls, fewer empty slots, staff in control.

Scope agreed ✓
02 · DesignBefore any code, we agree on the screens and the system. Trade-offs go on the table.
Staff dashboard · wireframe
Clinic · Today · 4 doctors search ⌕
Doctors
— Dr. Mehta
— Dr. Rao
— Dr. Iyer
— Dr. Shah
Live chats
↳ agent asks: which day? which doctor?
Free slots · Sat
10:1511:3016:00
Take over chat
System flow
Patient on chat→AI agent→Calendar
Reminders←Bookings DB⇢Staff
Decisions
  • WhatsApp, not a new app. Patients already use it.
  • Agent only offers free slots. It never promises what the calendar can't hold.
  • Staff override always wins. Any chat can be taken over in one tap.
03 · BuildThen I build it: the API, the AI agent and the dashboard. One engineer across every layer.
clinic-bookings
  • src/
  • agents/
  • booking.agent.ts
  • calendar/
  • calendar.service.ts
  • reminders/
  • reminders.cron.ts
  • dashboard/
  • today.page.ts
  • test/
  • booking.e2e.ts
booking.agent.ts · NestJS
@Injectable() export class BookingAgent {   @Tool('findSlots')   async findSlots(day, doctor) {     return this.calendar.free({ day, doctor });   }   @Tool('book')   async book(slot, patient) {     const held = await this.calendar.hold(slot);     await this.reminders.schedule(held, '-1d');     return held;   }   @Tool('handoff')   handoff(chat) { return this.staff.notify(chat); } }
Preview · patient chat
Namaste, Saturday ko Dr. Mehta mil sakte hain?
Saturday 11:30 aur 4:00 free hain. Kaunsa time chalega?
11:30 please
Booked ✓ Sat 11:30. Ek din pehle reminder aayega.
04 · ShipTested, containerised and deployed. The same steps run every time.
CommitTestsBuildDeployMonitor
TestsAll passingbooking · double-booking · takeover
Imageclinic-bot:1.0Docker · Node 20
RegionAWS Mumbaiclose to the clinic's patients
$ npm test ✓ booking flow · double-booking guard · staff takeover $ docker build -t clinic-bot:1.0 . ✓ image built $ deploy --env production ▸ rolling out to AWS… ✓ health checks passing · live
05 · LiveIt goes live and starts doing the work. Patients book themselves, staff see the day fill up.
Today's bookingsLIVE
0booked on chat
0reminders sent
0staff takeover
09:00Follow-up · Dr. Mehtavia chat
10:15New patient · Dr. Raovia chat
11:30Consultation · Dr. Mehtavia chat
16:00Check-up · Dr. Raostaff
Example data from the demo project
Reminder: Dr. Mehta kal 11:30 baje. Reply 1 to confirm.
1
Confirmed ✓ See you tomorrow.
06 · SkillsOne engineer across Interface, API, AI, Data and Cloud.
ONE ENGINEER / EVERY LAYERExperience, connected.
Interfacewhat people touch
TypeScript11–12yAngular9–10yReact7–8yNext.jsVue.jsTailwind
APIbusiness rules & services
Node.js9–10yNestJSSpring Boot.NET / C#RESTGraphQL
AIreasoning, retrieval, tools
GenAI3+yAgentic AIRAGMCPLangChainLangGraph
Datastate that holds
MongoDBPostgreSQLSQL ServerRedisRabbitMQVector DB
Cloudshipping & keeping it up
AWSAzureDockerKubernetesCI/CDCloudflare
The outcomeA clear brief. No guesswork.Scroll to continue
Back outside · your idea is next

Your project could be
next on this screen.

Start with a 60-minute diagnostic. You leave with priorities and a clear next step.

One product · every engineering layer

I build the systemsbehind the product.

Explore all 70 examples ↗
01 / 07Build the foundation

Platform engineering

A platform that helps every team ship.

  • Internal developer platform
  • Microservice scaffolding
  • Multi-tenant SaaS foundation
10 examples + case studies & portfolio
02 / 07Make AI do useful work

Agentic engineering

AI workflows with tools, limits and accountability.

  • Support triage agent
  • Grounded knowledge assistant
  • Planner–executor workflow
10 examples + case studies & portfolio
03 / 07Plan and test capacity

High-scale systems

Robust systems for demanding traffic.

  • Read-heavy public API
  • Payment event processing
  • High-volume webhook intake
10 examples + case studies & portfolio
04 / 07Find the weak points

System-design reviews

A review that turns uncertainty into a plan.

  • Service boundary review
  • Scalability review
  • Database design review
10 examples + case studies & portfolio
05 / 07Design the whole system

Solution architecture

Architecture your team can understand and build.

  • Multi-tenant product architecture
  • Agent-enabled operations
  • Event-driven integration
10 examples + case studies & portfolio
06 / 07Make integration predictable

SDK engineering

Developer tools people can trust in their code.

  • Typed API client SDK
  • Authentication SDK
  • Payments integration SDK
10 examples + case studies & portfolio
07 / 07Keep the product coherent

Design systems

A design system that survives product growth.

  • Token foundation
  • Accessible component library
  • Enterprise dashboard system
10 examples + case studies & portfolio
YOUR PRODUCTOne system.connected by design
01Platformfoundation
02Agentstool workflows
03Scalereliability
04Reviewtrade-offs
05Architectureboundaries
06SDKscontracts
07Designexperience
The foundation comes first.01 / 07
Scroll to assemble the systemExplore the engineering library ↗
Skills are connected. So is my work.

Follow a skill.
See the whole system.

Tap a neuron. Discover its professional context, every connected portfolio record and the architecture behind the idea.

ajay.neural / engineering map16 skills · live connections
AJAYEngineering
intelligence.
Connect the dots.
Select a skill to activate its connections
AI / connected context

Agentic AI

Tools, orchestration and accountable execution.

Illustrative architecture · not a delivery claim
A bounded agent execution loop

Bound time, cost and authority. Persist task state, inspect traces and use human approval for sensitive actions.

Agentic AI selected. Choose a connection to explore.Catalog labels preserved · architecture examples
Scroll to connect the engineering layers
Design for the workload

Millions of requests?
Give every request a path.

Follow one request through an illustrative architecture. Capacity comes from measured workloads, clear boundaries and recovery plans.

01Edgefilter & route
02APIvalidate & isolate
03Cachereuse the fast path
04Queuebuffer the burst
05Workersprocess & retry
06Datapersist safely
01 / 06

At the edge, filter and route before expensive work begins.

Scroll to follow the request · example architectureExplore scale & reliability ↗
Exploded view · the app you just watched

Open it up. Every layer is mine.

Scroll and the clinic app comes apart, layer by layer. Scroll back and it snaps together again.

Layer 1 / 5
Cloud
AWS
Docker
CI/CD
Mumbai
health ✓
logs
AI agent
agent
findSlots()book()RAG · FAQshandoff()
Database
bookings
patients
API GET/slots?day=sat POST/bookings POST/reminders POST/chats/:id/handoff 200 OK · 38 ms
UI
09:00chat
10:15chat
11:30chat
Take over
  • UIwhat people touch

    The staff dashboard and the patient chat experience.

    Angular9–10yReact7–8yNext.jsTypeScript11–12y
  • APIthe rules

    Bookings, reminders and handoffs, with the business rules enforced in one place.

    Node.js9–10yNestJSSpring Boot.NET / C#
  • Databasestate that holds

    Data and queues that never double-book a slot.

    MongoDBSQLRedisRabbitMQ
  • AI agentunderstands, decides, hands off

    An agent with tools and retrieval, under clear limits.

    GenAI3+yAgentic workflowsRAGMCP
  • Cloudruns it for real

    Containers, pipelines and monitoring, so it stays up.

    AWSDockerCI/CDCloud migrations
What the clinic got

Features unlock. Problems close.

Every feature is tied to a problem the client named on the first call.

  1. Booking on chat

    The front desk stops answering the same call all day.

  2. Day-before reminders

    Patients confirm with one reply, so fewer slots go empty.

  3. Replies in Hindi and English

    Questions at 11 PM still get a clear answer.

  4. Staff dashboard

    The whole day on one screen, and staff can take over any chat.

Clinic · chat

Dr. Rao Monday available hain?
Monday 10:15 aur 12:00 free hain.
10:15
Booked ✓ Mon 10:15 with Dr. Rao.
6:00
Sunday
CLINIC · NOWReminder: Dr. Rao kal 10:15 baje. Reply 1 to confirm, 2 to reschedule.

Clinic · chat

Kya Sunday ko clinic khula hai?
Sunday band hai. Monday 9 baje se slots hain, book karun?
Do you accept insurance?
Yes, the front desk will verify it at check-in.

Today · staff

09:00Follow-up · Dr. Mehta
10:15New patient · Dr. Rao
11:30Consultation · Dr. Mehta
16:00Check-up · Dr. Rao
Take over a chat
Patterns from client work

Work that holds up.

Explore case studies →
Portfolio case study

Knowledge, with a source

Teams had AI bolted on as isolated API calls. I replaced them with a governed flow where every answer traces back to a document.

Designed the AI flow, retrieval layers and service boundaries.
What's our refund window for annual plans?
30 days from purchase, pro-rated after that.12
[1] billing-policy.pdf · p.4   [2] faq/annual.md
Professional pattern

Automation, with a way back

Follow-ups, triage and approvals lived in tools that didn't talk. This pattern automates them and keeps a person able to step in.

Applied across client and product work.
New lead→Triage agent→CRM→You approve→Follow-up sent
Every automated step can be paused, reviewed or rolled back.
Independent initiative

A profile for the conversation

One generic CV can't speak to every role. I built a role-based profile system that now serves 94 targeted CV variants.

Designed the content model, architecture and implementation.
FullstackAngular · Node
ArchitectSystems · Cloud
AI EngineerRAG · Agents
Next slide

Your project goes here.

Start with a 60-minute diagnostic. You leave with priorities and a clear next step.

Book diagnostic · ₹2,499
your idea
→ sketched
→ built
→ live
The journey, fast-forwarded

From .NET to agents.

2012–26fourteen years, eight roles
    Hire · start small

    Pick how we start.

    Most projects begin with a short call. You leave it with a clear next step, whether or not you hire me afterwards.

    01 / 04
    Paid session
    ₹39930 min

    One focused working call on learning, product, architecture or AI.

    Book session
    02 / 04
    Best first step for businesses
    ₹2,49960 min

    B2B diagnostic. We look at a workflow, AI opportunity or platform problem and agree on priorities.

    Book diagnostic
    03 / 04
    Consulting & build
    Scoped

    Architecture, agentic AI, RAG and fullstack builds, scoped after the diagnostic.

    See services
    04 / 04
    Teams & companies
    Workshops

    Hands-on AI training so your engineers can run what gets built.

    Training
    Innovate with Ajay

    Don't just use AI.
    Build with it.

    innovatewithajay.com →
    LIVE EVERY SUNDAY

    Next session starts in

    0days
    00hrs
    00min

    English at 11 AM IST, Hinglish at 5 PM IST. Alternates between Build With Ajay (3 hrs, ₹399) and Clarity Sessions (1 hr, ₹99). Starts 11 October 2026.

    Reserve a seat
    Climb at your own pace
    Choose your next learning step.
    INNOVATEBuild
    with Ajay.
    Your path. Your pace.
    Contact · three taps

    Start the conversation.

    Answer three quick questions. The message gets written for you, and you send it on WhatsApp.

    Ajay PrajapatYour message goes straight to my WhatsApp.
    1 · You are

    Tell me a little
    about yourself.

    Choose the context that fits you.

    2 · You want

    What can we
    build together?

    Start with the kind of help you need.

    Keep it simple.

    A goal, a problem or a rough idea is enough to start.

    Your message
    Open in WhatsApp →
    WhatsApp +91 93105 39354Email ajayprajapat@live.com
    Have a project in mind?Let's talk