i am Rain
EN 中

Hi, I'm Rain

Forward Deployed Engineer AI Solutions Engineer Full-stack Engineer

Working with AI agents, I own the 0-to-1 delivery lifecycle—from requirement analysis to deploying live systems.Recently shipped an end-to-end ordering, payment, delivery, and order-tracking system used by real customers.Currently seeking Forward Deployed Engineer / AI Solutions Engineer / Full-stack Engineer roles.

View Projects → Get in touch

01 / Projects

Production workflows I shipped.

oppachikin

✅ Live · Taking orders

A full digital operations stack delivered 0→1 for a real paying client — a Minsk fried chicken shop: customer ordering on the outside, employee scheduling on the inside. End to end: I defined the rules and boundaries, then collaborated with AI agents to implement.

▸ Online ordering system

First launch in 5 weeks · 23 pages + 19 APIs · production-grade security

#Next.js#Supabase#Alfa Bank#SMS#Telegram Bot

▸ Scheduling bot

7×24 in production · employee self-service · monthly payroll auto-calc

#Python#Telegram Bot#Railway#Sheets

▸ Delivery pricing engine

Auto-samples courier fares every 10 min · priced from 2,592 samples · multi-exit dispatch with human fallback

#Vercel Cron#Yandex API#Supabase
See the full story →

02 / Experience

What I shipped.

A production system for a real paying client

PRODUCTION CUSTOMER SYSTEM

oppachikin

A production system for a real paying client

5 wks

shipped

1st

paid order

Deep Think

The most counter-intuitive lesson from shipping full-stack with AI agents: the more capable the AI agents are, the more important it is to plan first. Define the architecture, boundaries, and what AI agents can and cannot touch before writing the first line of code. Otherwise the implementation cost AI saves gets eaten by the cost of going in the wrong direction.

KA close-to-cash workflow automation

INTERNAL WORKFLOW AUTOMATION

SF Technology

KA close-to-cash workflow automation

6 reports

per month

1 day → 2 hrs

4× faster

Deep Think

The starting point for AI automation is not the technology—it is repeated exposure to the workflow. Only after operating a process end to end can you tell which steps should be automated and which should remain judgment-based.

Multi-platform data operations automation

DATA OPERATIONS AUTOMATION

Beyond Infinity Technology

Multi-platform data operations automation

+30%

efficiency

3 platforms

iOS / Google / PayPal

Deep Think

In scaled finance work, the real lever isn't doing each thing faster—it's abstracting the shared rules into a model. Build it once, use it 100 times.

Business-process diagnostics through audit and DD

BUSINESS PROCESS DIAGNOSTICS

KPMG Huazhen LLP

Business-process diagnostics through audit and DD

30+

subsidiaries consolidated

100+

major contracts reviewed

Deep Think

To deeply understand a company, you start with the financial statements—but the real insight is in the business flows of sales, procurement, production, and cash.

PRODUCTION CUSTOMER SYSTEM

oppachikin

A production system for a real paying client

5 wks

shipped

1st

paid order

For a local fried chicken shop in Minsk, I delivered a production operations system from 0 to 1: customer ordering, payment, delivery, and order tracking on the outside; Telegram scheduling, employee self-service, and monthly payroll automation on the inside. My role was to define the business rules, risk boundaries, and delivery rhythm, then collaborate with AI agents including Claude Code and Codex to implement.

  • Full-stack delivery: First launch in 5 weeks, covering 23 pages, 19 APIs, and the payment, notification, membership, and order-tracking flows. The system is live, payment tested, and has received a real paid order.
  • Data-driven pricing: A Vercel Cron job samples real-time Yandex courier fares for 8 fixed routes every 10 minutes, collecting 2,592 samples over about 5 days. The output: a weekday-hour peak grid × 10 distance bands fare table (weekend surcharge stacks, 12km+ goes manual), signed off by the owner and now live.
  • Multi-exit dispatch decisions: After discovering Yandex's scheduled-order quote parameter silently fails and the same route jumped over 11 BYN within 10 minutes in our samples, I designed dispatch as a multi-exit flow: re-quote at dispatch time + use the free claims/create as a true-price probe (accept is the only paid step) → auto-dispatch the normal case; route 12km+, anomalies, and failures to a human.
  • Production-grade security: A dedicated week of hardening—server-side amount validation, idempotent callbacks, webhook signature verification (constant-time comparison), tiered Redis rate limiting. I reviewed and debugged the critical path myself so code generated with AI agents could safely enter production.
  • Closed loop + internal ops: After the SMS interface went live, the first real paid order came in. The scheduling bot runs 7×24 (Python + Telegram + Railway): employees self-serve shifts, monthly payroll auto-calculated.

INTERNAL WORKFLOW AUTOMATION

SF Technology

KA close-to-cash workflow automation

6 reports

per month

1 day → 2 hrs

4× faster

Supported the downstream commercialization workflow of KA projects, covering project acceptance, invoicing ledgers, receivables registration, P&L tracking, and commercialization reporting. I automated and standardized 6 recurring monthly reports / ledgers, reducing effort from 1 day to 2 hours.

  • Acceptance + receivables workflow: Connected milestone acceptance tracking, document checks, receivables registration, and ledger maintenance into one repeatable process.
  • P&L and commercialization support: Scripted the monthly extraction, cleaning, and aggregation flow for KA project P&L, while supporting project-level cost confirmation for commercialization reporting.
  • Ledger and coordination structure: Standardized contract review, quotation review, contract filing, and project change tracking into structured ledgers instead of scattered Excel and email coordination.
  • Workflow-first automation mindset: The value was not just generating reports faster, but understanding the business flow first and deciding what should be automated versus what should remain judgment-based.

DATA OPERATIONS AUTOMATION

Beyond Infinity Technology

Multi-platform data operations automation

+30%

efficiency

3 platforms

iOS / Google / PayPal

Used VBA + Power Query to turn recurring work across revenue, ad spend, budgeting, and operating reports into reusable data models.

  • Multi-platform reconciliation: Built a VBA + Power Query pipeline to cleanse hundreds of raw iOS / Google / PayPal reports into one revenue reconciliation model, improving efficiency by 30%.
  • Ad spend aggregation: Monthly ad spend analysis across Google / TikTok / Facebook Ads, covering 100+ ad accounts with cross-platform cost attribution and expense audit.
  • Monthly cash budgeting: Collected and validated budgets across 10+ departments; built a budget-vs-actual variance model with ≤3% deviation alerts.
  • Operating reporting: Standardized monthly operating report templates with ROI / cost-structure visualization, supporting leadership decisions.

BUSINESS PROCESS DIAGNOSTICS

KPMG Huazhen LLP

Business-process diagnostics through audit and DD

30+

subsidiaries consolidated

100+

major contracts reviewed

Led annual audits and IPO-bound financial DD across PRC + HKFRS dual standards for HK / A-share listed clients.

  • Industry DD + financial validation: Designed interview lists and ran 10+ executive and business-unit interviews; cross-validated revenue recognition logic and cost structure; delivered 20+ pages of industry insight reports.
  • Business-process diagnostics: Reconstructed internal-control flowcharts for 4 core cycles (sales / procurement / production / cash) covering 20+ key controls; identified 15 control deficiencies and improved 5 customer risk mechanisms.
  • Listed-company audit: Led annual audits for multiple HK / A-share listed clients with PRC / HKFRS dual-standard consolidations (30+ subsidiaries); built bilingual workpaper templates that cut delivery cycle by 25%.
  • Project management: Tailored audit plans and key audit areas to industry characteristics; assigned and tracked team workload to ensure on-time delivery—this "look at flows before looking at numbers" training is the foundation of how I approach AI implementation today.

03 / About

Curious about the world.Genuine with people.Obsessed with making things happen.

Rain off the clock →
Curiosity

Driven by a natural curiosity about how the business world works.I thrive on diving into unfamiliar domains, getting my hands dirty on the front lines, and solving real-world problems.

Connection

Believer in authentic human connection.Whether founding the 'FUNergy' sports community or coaching 'Reset Life' simulation games, my goal is always to create a high-energy space where people unlock new possibilities together.

Action

Obsessed with turning ideas into reality.Coming from a financial audit background, I evolved from optimizing daily workflows to teaching myself AI-assisted coding and shipping full-stack systems. I don't just ideate—I learn, build, and execute until it runs seamlessly.

04 / Contact

Let's talk AI implementation.

I'm looking for Forward Deployed Engineer / AI Solutions Engineer / Full-stack Engineer roles, and I'm open to discussing real-world AI agents deployment scenarios.

Email →