Skip to content

Practice

AI voice agents and full-stack SaaS

Multi-tenant SaaS products and AI features that survive vendor changes, led by an engineer who ships them from the data model to the deployed interface.

Discuss your project

Problems we solve

  • Revenue lost to inbound calls nobody answers and follow-ups that never happen.
  • An AI feature tied to one vendor, so changing providers means a rewrite or downtime.
  • Leads, scheduling, billing and payouts that need to run in one system, across several roles.
  • Billing runs that could charge a customer twice when they are retried.
  • Leaderboards and dashboards that have to update live across more than one process.
  • A legacy system carrying documented correctness and security defects.

Capabilities

AI voice agents and full-stack SaaS

Full-stack SaaS products built end to end, with production AI features: real-time voice agents, LLM tool calling, transcript pipelines and vector search.

Led by Burraq Ur RehmanFull-Stack Engineer, AI Voice and SaaS Product Lead

AI in production

AI voice agents

Inbound call automation with real-time tool calling for bookings, and transcripts saved to the CRM.

  • Retell
  • x.AI Grok realtime
  • Twilio Media Streams
  • WebSockets
How it flows4 steps
  1. Inbound call
  2. Signature-verified webhook
  3. Voice agent with tool callsRetell · x.AI Grok realtime
  4. Booking and transcript saved

How we work on this

  1. 01

    First conversation

    Thirty minutes with the practice lead to understand the problem, the system it lives in and who depends on it.

    You get: A clear answer on whether we are a fit.

  2. 02

    Scoping

    The lead who will build the work writes down what will be built, how, and what could go wrong.

    You get: A written scope with approach, risks and price.

  3. 03

    Build in the open

    Weekly demos, written decisions and a shared board, so you always know what changed and why.

    You get: Working software every week, with a decision log.

  4. 04

    Ship and prove

    Tests run on the real database engine, the rollout is staged, and the handover is written down.

    You get: A staged release and a documented handover.

  5. 05

    Run and improve

    Optional retained support once the system is live, with agreed response times.

    You get: A support agreement, if you want one.

How we work in full

Proof

Automotive / AI SaaS

Automating Dealership Calls with a Multi-Tenant AI Voice Agent

Problem

Dealerships were losing revenue to missed inbound calls and dropped follow-ups.

What changed

An AI voice agent that answers inbound calls and books appointments through real-time tool calling, on two voice providers behind one switchable abstraction.

Result

Calls are handled automatically, and the platform can move between voice providers with zero downtime.

Delivered by Burraq as the engineer on Otto Revenue Engine, before Carpediemsoft.

  • Python
  • FastAPI
  • async SQLAlchemy
  • PostgreSQL
Read the case study
Sports / Real-time platform

Rewriting a Live-Scoring Tournament Platform

Problem

A legacy pigeon racing system carried 17 documented correctness and security defects.

What changed

A full rewrite with live leaderboards on PostgreSQL alone, bridging WebSocket updates across processes through LISTEN/NOTIFY.

128tests covering the rewritten platform

Result

A production platform covered by 128 tests, with a pipeline that typechecks, migrates, tests and audits dependencies before every deploy.

Delivered by Burraq on Pigeon Premier League, before Carpediemsoft.

  • TypeScript
  • NestJS 11
  • Prisma 6
  • PostgreSQL
Read the case study
All work

Led by

Burraq Ur Rehman

Full-Stack Engineer, AI Voice and SaaS Product Lead · AI voice agents and full-stack SaaS lead

Responsible for

Designs and ships full-stack SaaS products end to end, from the data model to the deployed interface, with production AI features such as real-time voice agents and LLM tool calling.

Experience

  • Built Otto Revenue Engine, a multi-tenant AI voice and CRM platform for automotive dealerships, on two voice providers behind one switchable abstraction.
  • Rebuilt Pigeon Premier League, a live-scoring tournament platform, replacing a legacy system that carried 17 documented correctness and security defects.
  • 5+ years as a full-stack developer at Xprolabs and Southville Solutions, where CI/CD cut deployment time by 40%.

Technologies

Languages
Ruby, Python, TypeScript, JavaScript, SQL, HTML5, CSS3
Frameworks
Ruby on Rails, FastAPI, NestJS, Next.js, Node.js, Express.js, Django, Flask, Celery, Pydantic v2, SQLModel
Frontend
React 18 & 19, Angular, Vite, TanStack Query, Redux, Context API, Tailwind CSS, shadcn/ui, Bootstrap, jQuery, AJAX
Backend
REST APIs, Microservices, Multi-Tenant Architecture, WebSockets, Socket.IO, Sidekiq, pg-boss, RBAC, JWT & JWKS Auth, Auth0, Devise, Passport, Stripe Billing, Third-Party Integrations
Databases
PostgreSQL, pgvector, SQLAlchemy 2 (async), Prisma, Alembic, Redis, NoSQL, Query & Schema Optimization
AI & ML
LLM Integration, Tool Calling, Real-Time Voice Agents, Retell, x.AI Grok Realtime, Twilio Media Streams, Transcript Processing, Embeddings & Vector Search, RAG, Prompt Design, scikit-learn Basics
Cloud & DevOps
AWS (EC2, S3, Lambda, RDS, CloudWatch), Vercel, Railway, Heroku, Docker, GitHub Actions, CI/CD, Nginx, PM2, Linux
Testing & Quality
pytest, RSpec, Playwright, Vitest, ruff, mypy, ESLint, RuboCop, Code Review
Ways of Working
Agile & Scrum (Scrum Master), Jira, Git, GitHub, GitLab, Bitbucket, Client Communication

Questions

Questions about this practice.

Can you put an AI voice agent on our phone line?

Yes. Inbound calls reach the voice agent through signature-verified telephony webhooks, it books appointments with real-time tool calls while the caller is on the line, and transcripts are saved to your CRM.

What if we want to change AI providers later?

Voice and LLM providers sit behind one abstraction. On a dealership voice platform, Retell and x.AI Grok realtime run behind the same switch, so the platform can move between them with zero downtime.

Do you work on Ruby on Rails?

Yes: new Rails products and feature work on existing Rails codebases, with access rules, RSpec coverage and RuboCop-enforced quality.

Can you rewrite a legacy system we have outgrown?

Yes. We audit it, fix root causes, add tests that prove the fixes, and ship through a CI-gated pipeline.

Have a problem like these? Tell us about it.

Discuss your projectSee our work