Skip to content
farshi.dev
Resume
Risograph portrait illustration of MD Salman Farshi.

Hello! I'm MD.

Salman Farshi

I build
Explore

Hello! I'm MD.

SalmanFarshi

Risograph portrait illustration of MD Salman Farshi.
I buildBackendsAgentsProducts

I build

Risograph portrait of Salman with warm halftone texture.

About me

IbuildAI-poweredproducts🤖andscalablebackendsystems.TypeScript,Node.js,Python,andGo.LLMs,RAG,real-timeservices,anddistributedarchitecture.Turningideas💡intoproductspeopleactuallyuse.

WhatI do

Risograph illustration of Salman working at a laptop.

Develop

Full-stack systems

Designing and shipping scalable web applications with clean APIs, robust backend architecture, and production-ready databases.
Node.jsNest.jsExpressMongoDBPrismaPostgreSQLRedisBullmq

Automate

AI-powered workflows

Building intelligent systems that combine LLMs, RAG pipelines, and agents to automate research, retrieval, and decision flows.
OpenAI APILangChainLangGraphRAGVector DBs

Engineer

Backend systems

Designing scalable backend architectures, APIs, and services that power real-world applications with reliability, performance, and maintainability.
Node.jsExpressMongoDBPostgreSQLRedisSystem Design

My career & experiences

Full Stack AI Developer

Core Devs LTD

NOW

Led architecture and backend development of a scalable SaaS platform, driving system design, APIs, and cloud infrastructure for reliable performance.

MERN Stack Developer

Soft Standard Solutions

2024

Optimized API response times by 40-60% through codebase optimization and efficient refactoring.

Full Stack Developer

CodeFoTech

2023

Developed and integrated scalable REST and GraphQL APIs using Node.js, Express, and MongoDB/PostgreSQL for high-performance backend operations.

Some of my work

01

Painless Agent

AI agent runtime

GoPostgreSQLpgvectorRedisSSELLMs
Risograph illustration of an autonomous agent runtime node graph with memory retrieval paths.
02

HireLoom AI

AI recruiting platform

ReactTypeScriptNestJSPrismaPostgreSQLBullMQGeminiS3
Illustration of the HireLoom AI recruiting platform with candidate-job matching signals.
03

Athlify

Sports & Community

NestJSPrismaPostgreSQLFirebase Cloud MessagingWebSockets
Risograph illustration of interface, design, and research symbols.
04

ProTrac

Factory operations

Django RESTNext.jsTypeScriptPostgreSQLRedis
Risograph illustration of Salman working at a laptop.
05

Oppora AI

AI sales SaaS

FastAPIPostgreSQLRedisCeleryLangChainGPT
Risograph illustration of connected code, browser, and API shapes.
RAG
AI Agents
Redis
PostgreSQL
Go
OpenAI
AWS
Docker
NestJS
WebRTC
MongoDB
LangChain
Socket.IO
TypeScript
BullMQ
Next.js
TensorFlow.js

Systems for AI workloads

Tools I reach for when the system has to hold.

Search systems

AI and Search

Search, retrieval, and agent loops that stay explainable.

LLMsRAGLangChainLangGraphOpenAI GPTGeminiVector SearchAgent WorkflowsAI-native SearchData Enrichment
Service edges

Backend and APIs

Production APIs and workers.

TypeScriptNode.jsNest.jsBullMQPythonFastAPIGoFiber
Reliability base

Data and Infrastructure

Storage, queues, deployments, and recovery paths.

PostgreSQLMongoDBRedisDockerGitHub ActionsLinuxProxmoxWindows ServerHyper-VActive Directory
Delivery surface

Frontend and Product

Interfaces and product flows shaped around real delivery.

Next.jsReactReduxTailwind CSSReact QueryZustandZodPlaywrightProduct PlanningStakeholder Collaboration

Field notes

All posts
Risograph illustration of books and reading notes.
01
Illustration of a distributed system handling one million requests per second through caching, load balancing, and asynchronous services.

2026

Let’s Handle 1 Million Requests per Second!

High-scale systems are not built by adding more servers. Reaching one million requests per second requires careful architecture, caching, asynchronous processing, and eliminating bottlenecks across the stack.

Read note
02

2026

AI Recruiting Systems Need Explainable Match Signals

Candidate-job matching becomes useful when the system can explain why a role fits, stream progress while analysis runs, and fall back gracefully when AI is unavailable.

Read note
03

2026

Building an Autonomous Agent Runtime in Go

Most agent frameworks hide the hard parts behind abstractions. Building Painless Agent meant confronting persistent memory, task planning, and long-running workflows without a safety net.

Read note