Тарас

Taras Sarakhman
Full-Stack Developer: React / Next.js / Node.js / TypeScript
Lviv, Ukraine | [відкрити контакти](див. вище в блоці «контактна інформація») | [відкрити контакти](див. вище в блоці «контактна інформація») | [відкрити контакти](див. вище в блоці «контактна інформація»)
SUMMARY
Full-stack developer with 4+ years of coding, the last 2+ in production teams across fintech, AI copilots, and real-time trading.
Most recently shipped an automated trading bot on Binance Futures: live market-data pipeline, risk limits, encrypted credential
storage, and an MCP server that lets AI clients drive the bot in plain language. Drawn to work where mistakes are expensive —
money flows, live data, LLM integrations.
TECHNICAL SKILLS
Programming Languages: JavaScript, TypeScript, HTML5, CSS3, SQL
Frameworks & Libraries: React, Next.js, Node.js, Express, NestJS, Capacitor, Redux (RTK Query), Zustand, Tailwind CSS,
shadcn/ui, Jest
DevOps & CI/CD: GitHub Actions, Docker, GCP, Vercel, App Store / Play Store release management
Developer Tools & Platforms: Git/GitHub, PostgreSQL, Prisma, Redis, Supabase, Firebase, Turborepo
Core Competencies: RESTful APIs, WebSockets, Microservices, JWT/OAuth, Performance Optimization, NFC integration,
i18n/RTL, BFF architecture, RAG/LLM, MCP servers, Agile/Scrum, Jira
EXPERIENCE
Full-Stack Developer (Contract), Private Financial Services Team (8 people, under NDA) — Feb 2026 – Jul 2026
Algorithmic Trading Platform
Stack: Next.js, TypeScript, NestJS (microservices), PostgreSQL, Prisma, Redis, Docker, GCP, GitHub Actions
- Built and deployed an automated trading bot on Binance Futures for a production pilot with 8 users, combining real-time market
data with ML-driven signals; executed ~17 trades per user daily, processing ~$36K in daily trading volume.
- Designed a confidence-based invalidation layer, discarding signals when model confidence dropped below a 70% threshold.
- Enforced a 1.5% max risk-per-trade cap on position sizing, with user-configurable margin, leverage, and concurrent-position
limits.
- Engineered a WebSocket ingestion pipeline (Redis + PostgreSQL) processing ~7K candles/day, handling timezone boundaries
and candle-aggregation edge cases; covered core logic with Jest.
- Shipped a type-safe trading dashboard (Next.js, shadcn/ui) with JWT-based auth and encrypted credential storage, isolating
trading accounts per user.
- Designed an MCP server exposing 8+ typed tools (positions, balance, risk settings) via HTTP/SSE, letting AI clients operate the
bot via natural language under a testnet-only, security-first design.
- Set up GitHub Actions CI/CD pipelines (automated tests, Docker image builds, deployment to GCP) and GCP Cloud Logging for
server monitoring across the trading platform services.
Mobile Ops Platform
Stack: React, Capacitor, Supabase Edge Functions
- Maintained a shared React (Capacitor) codebase across web admin and native iOS/Android apps; implemented NFC/AIOI
smart-card reading for asset tracking.
- Replaced polling with an event-driven push pipeline (Supabase Edge Functions), cutting manual dispatch time ~30%; owned
end-to-end App Store and Google Play releases for the production app.
- Onboarded an incoming developer during project handover: walked through the workflow and backend internals, reviewed and
approved their PRs.
Frontend Developer, Appexoft (Agency Contractor) — Sep 2024 – Feb 2026
Tax Management Platform
Stack: React, TypeScript, Mews PMS integration
- Developed a reusable search-select component with client-side multi-field filtering and sorting for Mews PMS records, replacing
manual list scanning.
- Added timezone-safe date handling and pre-flight validation for reservation queries, and built quarterly rental-income tax
reporting (per-property and portfolio-wide admin views) covering 23 properties and 200K+ utility records across 13 quarters.
- Fixed a Safari mobile infinite-load bug and a dark-theme flash-of-white-background render issue.
Tax & Accounting AI Copilot MVP
Stack: React, TypeScript, i18next, Tailwind CSS
- Architected the copilot frontend from scratch, proposing a Context + custom-hooks architecture designed to scale across
feature teams.
- Integrated a RAG-based LLM backend into a streaming chat interface with inline citations, folder-organized sessions, and tiered
usage limits.
- Implemented full Arabic RTL localization alongside English via i18next, and built a regex-based markdown transpiler rendering
LLM output as interactive UI.
Restaurant CRM System
Stack: Next.js, TypeScript, Turborepo, Kinde SSO, Tailwind CSS, shadcn/ui
- Implemented a Next.js API-route BFF proxying legacy and new backend APIs for zero-downtime migration, plus tenant-scoped
route authorization with next-intl locale routing.
- Wrote project documentation and led a one-week handover, supporting the incoming team on process and
client-communication questions.
Freelance Full-Stack Developer — Aug 2022 – Aug 2024
- Crypto Trading Platform (Next.js, WebSockets): eliminated race conditions in real-time order-book/price-ticker feeds and
performed performance optimization of chart/ticker rendering to sustain 60 FPS during volatile markets.
EDUCATION
Lviv Polytechnic National University, B.S. in Software Engineering — Expected 2028
LANGUAGES: English (B2), Ukrainian (Native)