Open to full-time roles & consulting

I build like it's
my own company

Full-stack & AI engineer with 8+ years building and operating production software, now shipping AI products end to end: realtime voice agents, LLM agent pipelines, and the self-hosted platform they run on. Before that: four years as the sole developer behind a live marketplace SaaS, owning everything from the database schema to the deploy pipeline.

8+
Years in Production
4
Years Owning One Platform
3
AI Products Shipped in 2026
View My Work Book a 30-Min Call
Sandeep Singh - Full-Stack & AI Engineer

My Story

I've been the stakeholder, the founder, and the engineer. That order matters.

From the business side in

I didn't start as a programmer. I started running IT for a hotel in India — replacing paper billing with cloud POS, putting rooms online. That's where I learned the lesson that still shapes how I work: technology only matters when it makes the business run better.

Then I co-founded a software company, building products for healthcare and media clients. As a founder you don't get to say "not my job" — you own the outcome, the budget, and the relationship. I've carried that ownership into every team since.

Most recently I spent four years as the principal developer at Salle Privée, a Montreal event-venue marketplace — designing the architecture, building the product, and running it in production. When you're the one answering for the whole system, you learn to build what's needed, not just what's asked.

Since 2026 I've been shipping AI products end to end for early-stage founders: a medical companion app with a realtime voice agent, an AI voice-operations platform for property management, and a multi-tenant outreach CRM with its own voice agent — designed, built, and operated by me, with a fleet of coding agents I orchestrate through my own task-broker API. Outcome-first by habit: pick the problem that moves the product, ship it, instrument it, and let the numbers decide what's next.

Founder's Mindset

I've run a company. I think about cost, risk, and revenue — not just code.

End-to-End Ownership

From database schema design to CI/CD pipelines to production incident response — one person accountable for the whole system.

Technical Depth

Python/FastAPI, Node.js/TypeScript, Elixir/Phoenix, React/Next.js, PostgreSQL + pgvector, LangGraph, Pipecat, Langfuse, Docker, Cloudflare, AWS

How I Work

Four principles from a decade of owning production systems

What's needed, not just what's asked

When the feature you asked for is the wrong one, I'll say so — and show you what I'd build instead and why. You get an honest engineering partner, not a yes-machine working through tickets.

Your vision becomes my vision

I don't work for companies, I work as part of them. I learn the business model, sit with the people who feel the pain, and make technical calls that serve where the company is going — not just the sprint in front of me.

Own it end to end

Schema design, payments, real-time features, CI/CD, monitoring, the incident at an inconvenient hour — I take responsibility for the whole system, not a slice of it. Software I ship is software I operate.

Stakeholders stay in the loop

Founders and non-technical partners always know where things stand, what things cost, and what the trade-offs are — in plain language. Good engineering includes making sure the people around it are doing well.

Featured Work

Real projects solving real business problems

NOW — LEAD ENGINEER, AI PRODUCTS

Three AI Products for Early-Stage Founders

Design, ship, and operate production AI products end to end for founders who bring me their idea, using a fleet of coding agents (Claude Code, Codex, pi, grok) that I orchestrate through a task-broker API and shared memory.

Medical companion app

Multilingual voice check-ins, medication tracking, visit recording with live translation, AI briefs. FastAPI backend, Pipecat realtime voice agent (Azure Speech, GPT-4o on Azure OpenAI), LangGraph StateGraph for post-visit processing, PostgreSQL + pgvector with row-level security for caregiver sharing, a safety supervisor layer with an escalation ladder, and an offline reviewer that re-reads every transcript. Expo/React Native app with WebRTC voice. Every LLM and voice call traced in Langfuse.

AI voice-operations platform for property management

A phone agent over Twilio Media Streams with per-call runtime routing and automatic failover between Gemini Live speech-to-speech and a cascade pipeline (Deepgram Nova-3 → Llama-3.3-70b on Groq → Deepgram Aura); agent tool calls hit a C#/.NET API that owns buildings, residents, tickets, escalation, billing, and audit logs.

Multi-tenant outreach CRM

WhatsApp, email, meeting booking, and voice campaigns. Built the inbound/outbound voice agent (Plivo, Deepgram, OpenRouter) and the campaign, inbox, and meeting-booking workflows around it (Node/Express, BullMQ, MongoDB, React); consolidated all LLM traffic onto one vendor to cut cost and operational surface.

Python / FastAPI Node.js / TypeScript LangGraph Pipecat Twilio / Plivo Deepgram Azure OpenAI / Speech pgvector Langfuse React Native
3 Products
Voice agents, LLM pipelines, agent-operated infra
End to End
Idea to production, operated by me
Traced
Every LLM and voice call in Langfuse and Sentry
2022–2026 — 4 YEARS AS PRINCIPAL DEVELOPER

Salle Privée — Event-Venue Marketplace

A production SaaS platform connecting event organizers with venue providers in Montreal. I was the principal developer for four years — from architecture through production monitoring.

View Live Platform → 🚀

The Challenge

A two-sided marketplace needs everything a big product team builds — search, quoting, real-time messaging, payments, admin tooling — but here it all had to be designed, shipped, and kept running reliably by one accountable engineer.

What I Built

An Elixir/Phoenix LiveView core on PostgreSQL with a 44-table domain model, plus Node.js microservices for PDF generation and image processing. Stripe payments with refund workflows, real-time messaging over Phoenix Channels, and multi-channel notifications (email, SMS, in-app) on Oban background jobs. A two-way QuickBooks Online integration kept customers, products and invoices in sync with the company's books, with installment payments recorded as linked payment receipts. I also owned the infrastructure: Docker, GitHub Actions CI/CD, AWS, and incident response with Sentry.

The Scope

44
Table Domain Model
3
Notification Channels
4 yrs
Sole Technical Owner
24/7
Live in Production
Elixir / Phoenix LiveView PostgreSQL Node.js Stripe AWS Docker
4 Years
One platform, owned end to end
Full Stack
Architecture to operations
Real-Time
Messaging, payments & search
CO-FOUNDER

ZENI Applied Sciences

Co-founded a software firm serving healthcare and media clients. We shipped a real-time news aggregator app and an operations platform for a multi-location clinic franchise — scheduling, workflows, and management in one place.

Real-time news app on Flutter & Firebase
Clinic-franchise web platform
Delivery across small client teams
Flutter Firebase Web
CLIENT PROJECT

Levart

A workflow automation platform for business process orchestration, built in Ruby on Rails — turning manual, error-prone processes into structured flows.

Task automation engine
Approval flows & process management
Backend logic & data modeling
Ruby on Rails Automation
ALWAYS RUNNING

Agent-Operated Platform

The page you're reading is deployed from a 3-node self-hosted platform I run myself — 16 backends behind Cloudflare Tunnels, Pages, R2, and Workers, with shared Postgres/Mongo/Redis — operated largely by agents. I don't just talk about operations — I live with them.

Telegram-to-task-broker API for coding agents to claim and report work
Cross-agent memory store and MCP tool servers in agent workflows
Fleet health agent with alerting; push-to-deploy CI for 16 apps
Docker Cloudflare Claude Code / Codex MCP Linux

Want to go deeper on any of these?

Ask Me About the Details
View Public Repos on GitHub

What People Say

Recommendations from colleagues and clients

GP
Guillaume Paul-Simoncelli
Président et fondateur @ Salle Privée
May 2026 • Letter of recommendation • Managed Sandeep directly for four years

"For most of that journey, Sandeep was our lead developer and frankly, the technical backbone of the product. The platform he designed and maintained handled the full user journey: venue discovery, quote requests, real-time messaging, Stripe-powered payments, admin dashboards, and a multi-channel notification system spanning email, SMS, and in-app."

"Beyond the technical skills, what stands out to me about Sandeep is his reliability. When something broke in production on a Sunday evening, he was there. When a complex feature needed to ship for a critical client, he found a way. He took ownership of mission-critical systems for years without us ever having to second-guess the quality or stability of his work. For a founder, that kind of trust in your technical lead is genuinely rare and incredibly valuable."

"I recommend Sandeep without reservation to any organization looking for a senior full-stack developer with real DevOps depth."

EB
Eric Benoit
Consultant informatique @ Cyber Atlantide
March 2022
"I'm very happy to give a recommendation to Sandeep. He is highly efficient and focused developer. As a technician myself I really like working with him. He very good at finding solutions to any complex problems. His passion to satisfy the client makes the understanding easier to get for me and him. Sandeep knows applications development very well and his skills are helped me to organize my grow."
RK
Rahul Kumar
Product Director @ Revivo Technologies
May 2020 • Managed Sandeep directly
"I am happy to write recommendation for Sandeep. He is highly productive and focused person. As a team mate he is fun to work with. He is innovative in nature and and good at finding solutions to complex problems. His passion for computer and technology makes things easier to do. Sandeep knows web-development and mobile development very well. His managerial skills are stunning, and helped the organisation to grow."
View More Recommendations on LinkedIn

The Journey

From a hotel front desk in India to shipping AI products in Montreal

2015 — Sasaram, India

IT Manager, Maurya Royal

My first job was making technology serve a business, not the other way around: replacing manual billing with cloud POS and putting the hotel's rooms online. I've thought like an operator ever since.

2017 — Noida, India

Co-Founder, ZENI Applied Sciences

Started a software company building web and mobile products for healthcare and media clients. Three years of owning outcomes — code, deadlines, budgets, and relationships.

2020 — Noida, India

Full-Stack Developer, Anand Soft World

Client web applications across the full product lifecycle — frontend, backend, and the REST APIs connecting them to the outside world.

2022 — Montreal, Canada

New country, new degree, one platform

Moved to Montreal, completed a DEC in Computer Science at Collège LaSalle, and joined Salle Privée — where I spent four years (May 2022 – May 2026) as the principal developer of their event-venue marketplace, owning it from schema design to production incidents.

2026 — Montreal, Canada

Lead Engineer, AI Products (independent)

Three founder-led products, built and operated end to end: a medical companion app with a realtime voice agent and LangGraph pipeline, an AI voice-operations platform for property management with failover between two voice runtimes, and a multi-tenant outreach CRM with its own voice agent — all running on a 3-node platform operated largely by coding agents I orchestrate.

Now

Looking for the next company to build with

Eight years in production, four owning one platform, a year shipping AI products alone. I'm looking for the next team whose vision I can make my own — as a senior full-stack & AI engineer or a hands-on technical partner.

Tech Stack

Backend

  • Python / FastAPI
  • Node.js / TypeScript / Express
  • Elixir / Phoenix (LiveView, Oban)
  • Ruby on Rails, C# / .NET
  • REST / GraphQL APIs & WebSockets

Frontend

  • React / Next.js (TypeScript)
  • Expo / React Native (WebRTC)
  • Phoenix LiveView
  • Svelte, Tailwind CSS

AI & Agents

  • LangGraph, Pipecat agent pipelines
  • Realtime voice: Twilio / Plivo media streams, Deepgram, Azure Speech, Gemini Live
  • OpenAI / Azure OpenAI, Anthropic, Groq, OpenRouter
  • pgvector retrieval, MCP tool integration, Langfuse tracing
  • Agentic development: Claude Code, Codex, pi

Data, Cloud & Ops

  • PostgreSQL (+pgvector, row-level security), MongoDB, Redis
  • BullMQ / arq / Oban job queues
  • Docker, Cloudflare (Tunnels, Pages, R2, Workers)
  • AWS (EC2, RDS, S3, SES, Lambda), Azure OpenAI / Speech
  • GitHub Actions, Sentry, Linux

Let's Build Together

If you're hiring a senior full-stack & AI engineer who'll treat your company like their own — or you need someone to take an AI product from idea to production and operate it — I'd love to hear what you're building.

Email

mail@hellosandeep.com

GitHub

@sandeep2194

Location

Montreal, Canada

Available for:

Full-Time Roles Contract Development Technical Consulting Architecture & Code Review Mentoring

Based in Montreal • Remote-friendly • On-site in the Montreal area