Samarpit Santoki

Building AI Platforms.

I design and build the systems underneath AI products — communication infrastructure, realtime voice agents, workflow orchestration, and memory layers that power real-world operations.

Explore My Work Download Resume Get In Touch

Open to AI platform, backend infrastructure, and product engineering roles.

Scroll

Platform Case Studies

Deep dives into platform systems I've designed and built — architecture, trade-offs, failure modes, and lessons learned.

Unified Communication Infrastructure

CommAgent

Centralized communication platform replacing fragmented legacy systems across Email, SMS, WhatsApp, and Voice.

150k+

conversations / month

100+

customer environments

Case Study

AI Voice Platform

DINA

Realtime AI voice agents for customer onboarding and operations.

Inbound + outbound

voice workflows

Low-latency

streaming pipeline (< 1s target)

Case Study

Agentic Workflow Platform

Workflow Platform

Workflow infrastructure for customer-specific AI automations, agent delegation, and multi-channel execution.

AI Context Engine

Context Engine

Shared memory and context layer for AI agents and platform services.

Persistent

memory across sessions

Ranked

retrieval within token budgets

Architecture

What I'm Building

Currently on my desk.

AI Context Engine

Active

Memory and context infrastructure for AI agents — persistent memory, ranked retrieval, relevance decay, and shared context for multi-agent coordination.

Agentic Workflow Systems

Active

Systems where AI agents execute tasks, delegate work, call tools, and communicate across channels through workflow orchestration.

Communication Platform

Active

Evolving unified communication infrastructure — deeper workflow integration, better routing, richer analytics, and stronger observability across channels.

Tradex

Personal

A derivatives trading platform built in Go and Rust — matching engine, risk checks, market data, wallet, and settlement infrastructure.

About Me

I build platform software that lets engineers and businesses move faster. Recently, that has meant working on unified communication infrastructure, real-time voice AI, agentic workflow systems, and memory layers for AI agents. I care about clean contracts, failure modes, observability, and systems that survive real production usage — not just demos.