rahul gupta — ~/portfolio

// about.md

I'm a senior software engineer who lives in the uncomfortable middle: product UX on one side, pipelines, infra, and quantitative logic on the other. I enjoy owning that gap end-to-end — from the Docker image to the button the user clicks.

Today I build high-availability auto-trading systems at QuantDirection — CI/CD from scratch, webhook-driven pipelines, centralised logging. Before that, four years at Nexsales shipping an Artificial Neural Network with 95%+ accuracy, rolling out single-OAuth across 6+ products, and running GCP + AWS deploys.

Tools I reach for without thinking: Node.js, Python, Go, Docker, GCP/AWS, Next.js, PostgreSQL. Also TensorFlow when the problem asks for it. I care about small surface area, short feedback loops, and code that a stranger can read on a Monday morning.

// experience — most recent first

Jan 2025 — present

Senior Software Engineer

QuantDirection (via Cloudstaff LLP) — California, US · remote

Onboarded by reading the codebase end-to-end and producing a documentation knowledge base for the team.
Refactored a legacy codebase and resolved long-standing bugs, improving maintainability as a foundation for scale.
Built CI/CD pipelines from scratch, enabling reliable, rapid deployments where none previously existed.
Designed auto-trading pipelines using webhooks + third-party integrations once infra was stabilised.
Set up centralised logging and monitoring for full visibility into high-availability trading operations.

Node.jsPythonDockerCI/CDWebhooksGCP

Oct 2022 — Jan 2025

Software Engineer → Senior Software Engineer

Nexsales Solutions Pvt Ltd — Mumbai, India

Developed an Artificial Neural Network in TensorFlow with 95%+ accuracy; exported from Python to JavaScript for runtime performance.
Implemented single OAuth via Auth0 across all products — cut sign-up drop-offs by 20% and security issues by 30%.
Revitalised a JavaScript scraping tool (+40% speed); automated keyword scraping via shell scripts.
Collaborated with architects to build scalable infra for a lead-gen platform serving thousands of users.
Managed CI/CD on GCP, used AWS for scalable cloud; contributed to 6+ products across varied stacks.

Node.jsTensorFlowNext.jsVueAuth0GCPAWS

Jan 2021 — Oct 2022

Full Stack Web Developer

Saryu Business Solutions — Mumbai, India

Led development of 6+ full-stack web apps in PHP + MySQL, boosting user engagement by up to 25%.
Built a static-site generator with custom templates to productise delivery — cut cost + time by 40% across projects.

PHPMySQLNode.jsPython

rahul@amrahul.in:~$ ls -lah projects/

// selected work — 5 of many

01auto-trading-pipeline/● live · high-availability

2025 — present · QuantDirection

2025 · quant · backend + infra

Auto-trading pipeline at QuantDirection.

Webhook-driven pipelines that wire third-party signals into live trading flows. Stabilised a legacy codebase, built CI/CD from scratch, upgraded core dependencies, and set up centralised logging for full visibility into high-availability operations.

Node.jsPythonWebhooksCloud Infra

confidential — details on request

pipeline · ops view

trading-engine · produptime 99.98%

webhooks ingested12.4k / d

deploy frequency4.2 / wk

p95 latency140ms

logs centralised✓

CI/CD

0 → 100%

stack

Node · Python

availability

02neural-net-classifier/● shipped

2023 — 2024 · Nexsales

2023 · ML · Python → JS

Artificial Neural Network — 89% accuracy on data-gap filling.

Trained an ANN in TensorFlow (Python) to fill missing data gaps in a lead-gen dataset using patterns from past records. Exported the model to JavaScript for runtime performance. Internal tool used by the Nexsales team.

TensorFlowPythonJavaScriptNode.js

internal — Nexsales

   ┌────────────────────────────┐
   │  model · ann_v3.h5         │
   │  ───────────────────────── │
   │  input        ▓▓▓▓▓▓▓▓     │
   │  hidden · 1   ▓▓▓▓▓▓       │
   │  hidden · 2   ▓▓▓▓         │
   │  output       ▓▓           │
   │                            │
   │  accuracy   ███████░░  89% │
   │  exported  → tfjs · prod   │
   └────────────────────────────┘

accuracy

89%

runtime

py → js

task

data-gap fill

03single-oauth-auth0/● shipped

2023 · Nexsales

2023 · platform · auth

Single OAuth across 6+ products via Auth0.

Replaced a scattered auth surface with single sign-on across every Nexsales product. Cut sign-up drop-offs by 20% and security issues by 30%. Also revitalised a JS scraping tool — +40% speed — and automated keyword scraping via shell scripts.

Auth0Node.jsNext.jsVue

internal — platform-wide

   products/
   ├── lead-gen        ┐
   ├── prospector      │
   ├── enrichment      │──►  auth0
   ├── analytics       │       │
   ├── admin-console   │       ▼
   └── scraper-tool    ┘    single
                            sign-on

   drop-off  ↓ 20%
   security  ↑ 30%

drop-off

−20%

sec issues

−30%

products

04static-site-generator/● shipped

2021 — 2022 · Saryu

2021 · productisation · PHP + MySQL

Static-site generator — productising web delivery.

Built a templating engine + generator at Saryu Business Solutions that productised static website delivery. Cut dev cost + time by 40% across 6+ client projects, with user engagement wins up to 25%.

PHPMySQLTemplatesShell

client work — Saryu

   $ ./ssg build --theme=agency
   █████████████████████ 100%

   ✔ templates resolved
   ✔ pages rendered       42
   ✔ assets optimised    118
   ✔ sitemap · robots
   ✔ ci · ftp deploy

   cost    ↓ 40%
   time    ↓ 40%
   clients   6+

cost ↓

40%

engagement ↑

25%

apps

05auto-research-engine/● live · private

2025 — present · personal

2025 · AI · Python + Claude + EC2

Autonomous research agent with self-healing loops.

Designed a prompt + code architecture that runs multi-iteration autonomous research via the Claude API. Includes robust self-healing mechanisms that detect and recover from stalls mid-run. Deployed on AWS EC2 with a web dashboard — users can kick off research iterations and monitor progress live.

Claude APIPythonAWS EC2Dashboard

confidential — details on request

research-engine · dashboard

autoresearch · ec2 · prod● running

iteration7 / 10

sources fetched148

self-heals2

agentclaude api

model

Claude API

infra

AWS EC2

resilience

self-healing

// stack — tools I reach for without thinking

backend/

Node.js ●●●●●
Python ●●●●○
Go ●●●○○
PHP ●●●●○
Shell / Bash ●●●●○

frontend/

React.js ●●●●●
Next.js ●●●●●
Vue ●●●●○
TypeScript ●●●●○
API design ●●●●●

devops-cloud/

Docker ●●●●○
GCP ●●●●○
AWS ●●●●○
CI/CD ●●●●●
Monitoring ●●●●○

data-ai/

PostgreSQL ●●●●○
MongoDB ●●●●○
TensorFlow ●●●●○
ML / ANN ●●●●○
Auth0 ●●●●●

// identity

Rahul Gupta.