I build systems that
work while I sleep.

Gulfam Khan — Full Stack Engineer, 3 years in production. I build AI-integrated systems with reliability discipline: LLM scoring with deterministic guardrails, pipelines that survive bad inputs, and dashboards nine business teams run on daily. React, Next.js, Node.js, TypeScript, across five production databases.

operational·3 autonomous services on GCP Cloud Run·~2,800 documents AI-scored daily·unattended since May 2026

3 yrs

production experience

200+

backend modules maintained

5

databases in production

9

business-team dashboards

work

Shipped & running

Market Analyser

source ↗

AI-powered investment research platform: 42 REST APIs scoring 15+ US equities across 18 technical indicators with live data from Yahoo Finance, FRED, and Finnhub. Market-regime detection, VaR and Monte Carlo risk analytics, and Claude-generated bull/base/bear scenario analysis — structured AI reasoning over noisy real-world data.

Next.jsExpressTypeScriptClaude API

42 endpoints · 18 indicators · 3 market data providers

Enterprise Dashboards at Mylo

internal — ask me about it

Analytics platform nine business teams use daily — warehouse, finance, logistics, marketing, category, HR. 50+ dashboard pages on React 18 + TypeScript over a 200-module Node.js/NestJS backend, querying MySQL, MongoDB, Redis, BigQuery, and Databricks. Includes Claude-assisted analytics tooling for non-technical teams.

React 18NestJSMySQLBigQueryDatabricksRedis

9 teams · 50+ pages · 230+ components · 5 database technologies

Autonomous Intelligence Pipeline

source ↗

A two-service system that ingests from email (OAuth), scrapers, and third-party APIs, then ranks everything with Claude — LLM judgment bounded by deterministic guardrails, so a confident model can never override a hard rule. Built to run unattended: per-item failure isolation, lenient parsers for truncated LLM output, deadline budgets, secrets in Secret Manager, least-privilege service accounts. It has run every morning since May — including the days its inputs turned hostile.

Node.jsGCP Cloud RunMongoDBClaude APIDocker

2 services · Cloud Run asia-south1 · daily crons · unattended since May 2026

experience

Software Developer · Mylo

July 2023 — present · full stack

skills

Toolbox

AI & LLM

Claude APILLM scoring pipelinesPrompt engineering + deterministic guardrailsAI-assisted development

Frontend

ReactNext.jsTypeScriptReduxTailwind CSS

Backend

Node.jsExpressNestJSREST designJWT / OAuth

Databases

MySQLMongoDBRedisBigQueryDatabricks

DevOps

DockerGCP (Cloud Run, Secret Manager, Scheduler)GitHub ActionsMongoDB Atlas

contact

Talk systems with me.

I like conversations about making AI reliable in production — guardrails around LLM judgment, pipelines that degrade gracefully, dashboards people actually use. Building something in that space, or disagree with something I wrote? Say hi.