Msh.MAHDIEHSHAVANDI@YMAIL.COM
Mahdieh Shavandi
35.6892° N · 51.3890° E
TEHRAN / WORLDWIDE

MAHDIEHSHAVANDI

AI SOFTWARE ENGINEER · PRODUCT BUILDER

I BUILD AI PRODUCTS.

From first commit to production.
Applied AI, reliable software, and everything it takes to ship.

OPEN TO REMOTE ROLES & RELOCATION
SCROLL THE STORY

THE JOURNEY

0+

YEARS BUILDING SOFTWARE

DISCOVER

Start with a real problem.

From robot controllers to mobile products across France, Germany and Canada. Now building and shipping AI products end to end.

IDEA → ARCHITECTURE → PRODUCTION

HOW THE WORK ACTUALLY HAPPENS

AT THE MACHINE.

› defining the problem before writing a lineTypeScript · Python · curiosity

01

THINK

Understand the people, the constraints, and the problem worth solving.

02

DESIGN

Shape the system: data, APIs, tenancy, model behavior, and risk.

03

BUILD

Turn the decision into software: mobile, web, backend, automation, AI.

04

SHIP

Production is the real test. Deploy, observe, recover, and own delivery.

05

EVOLVE

Listen to real usage. Improve performance, reliability, and the product.

EXPERIENCE

ACROSS
INDUSTRIES.

Different products. Different constraints.
The same commitment to making them work.

FIELD NOTES / ENGINEERINGCLASSIFY → GENERATE → REVIEW

APPLIED AI

01 / 06

CASE STUDIES

PROOF,
NOT PROMISES.

01

MINIKASKO ↗

From an idea to a live, LLM-powered SaaS.

2026 — PRESENT
MINIKASKO / BRAND PROFILE / ENGLISH DESKTOP PREVIEW

An AI product for Instagram businesses that replies to customer comments in their own tone. I founded MiniKasko and owned the full journey from architecture to launch.

THE ENGINEERING

5,719PUBLIC LABELED EXAMPLES
243PROJECT-SPECIFIC EXAMPLES
INT8LOCAL ONNX INFERENCE

THE CHALLENGE

Make customer replies relevant to each account’s tone and post context, while handling Persian abuse, provider failures, and tenant isolation.

WHAT I DID

  • Built TypeScript, Fastify, PostgreSQL and Next.js services, authentication, payments, OAuth and webhooks.
  • Integrated LLM classification and generation with retries, fallback and human review.
  • Fine-tuned ParsBERT and exported it to ONNX INT8 for local inference.

THE RESULT

  • Launched a multi-tenant AI SaaS
  • PostgreSQL RLS tenant isolation
  • Local classifier with LLM fallback
  • Vitest tests gated in GitHub Actions

TYPESCRIPT / FASTIFY / POSTGRESQL / NEXT.JS / PARSBERT / ONNX / DOCKER

02

THE CAREER AGENT ↗

Turning a conversation into opportunities and a tailored resume.

MAR — SEP 2026
01CANDIDATE
PROFILEPersian / English
→
02JOB
SEARCHSerpAPI · Google Jobs
→
03LLM
MATCHINGPrimary + fallback
→
04TAILORED
RESUMEDOCX · Telegram · Bale
A NEXT STEP.
NOT ANOTHER SEARCH.WORKFLOW VISUALIZATION / SELF-HOSTED N8N

A public bilingual job-search and resume agent. It retains candidate context, finds relevant roles, and delivers resumes tailored to the opportunity.

THE CHALLENGE

Connect job search, candidate profiles and resume generation across languages, countries and a stateful conversation.

WHAT I DID

  • Connected n8n, LLM APIs, SerpAPI and Google Docs/Drive.
  • Retained profiles, language preferences and progress in Data Tables.
  • Deployed with Docker on a VPS and resolved search, language and document delivery failures.

THE RESULT

  • Live public Telegram bot
  • Persian and English support
  • Country-specific job matching
  • Tailored DOCX delivery

N8N / LLM APIS / SERPAPI / GOOGLE DOCS / TELEGRAM / DOCKER

03

SANTAFOO ↗

Making a production mobile app faster, more reliable, and easier to ship.

FEB 2022 — JUN 2026

PRODUCTION / MOBILE ENGINEERING

BEFORE100%
AFTER19%
81%FASTER STARTUP
Official SANTAFOO iPhone App Store preview showing the home screen and grocery categories
Official SANTAFOO iPhone App Store preview showing grocery products and adding items to the cart
OFFICIAL APP STORE PREVIEWS · STARTUP RESULTS FROM RESUME

Four years owning the React Native app at a French startup. Features, architecture, iOS and Android releases, staged rollouts, monitoring, and hotfixes.

MEASURABLE IMPROVEMENT

81%STARTUP IMPROVEMENT
~70%FEWER REPORTED FRONT-END BUGS
~80%UP TO / SCROLL & CART RESPONSIVENESS

THE CHALLENGE

Improve everyday performance and reliability while continuing to deliver features and own the mobile roadmap through a leadership transition.

WHAT I DID

  • Profiled bundle loading, rendering and caching.
  • Used Sentry to prioritize issues and prevent regressions.
  • Coordinated delivery with engineering, marketing and operations.

THE RESULT

  • Faster app startup
  • Substantially fewer reported bugs
  • More responsive shopping flows
  • End-to-end release ownership

REACT NATIVE / TYPESCRIPT / SENTRY / IOS / ANDROID / RELEASE OPERATIONS

MORE ALONG THE WAY

IntSoftReact Native Developer2021 — 2022 +

Germany · Part-time. Built a social-media app from scratch, establishing shared components and reusable UI patterns.

WalletifaiReact Native Developer2021 — 2022 +

Canada · Part-time. Built bill-management features, added Jest regression coverage, mentored teammates and contributed code reviews.

MooTanRooReact Native Developer2021 +

Tehran. Sole mobile engineer on a new TypeScript app; introduced Detox end-to-end testing for critical flows.

Earlier teamsMobile & Robotics2017 — 2020 +

Built mobile products at FaraRah, Mopon, Dadekavan and SAAP. Led and mentored developers at SAAP, and built Bluetooth and USB software for 8/16-DOF robots at KebriaCo.

BEYOND THE SCREEN

SOMETIMES
I BUILD
IN THE REAL WORLD.

Robot controllers. A chocolate 3D printer. A 6-DOF robot path-control thesis. A different playground for the same engineering mindset.

EDIPRINTS · CO-FOUNDER & CTO · 2018–2021BSC MECHANICAL ENGINEERING · 2017IMAM KHOMEINI INTERNATIONAL UNIVERSITY

CONTACT

WHAT SHOULD WE
BUILD
NEXT NOW?

AI products. Thoughtful engineering. Real ownership.
Open to remote roles, contract work and relocation.

mahdiehshavandi@ymail.com
LINKEDIN ↗GITHUB ↗RESUME ↓
SCENE 01