Arkadiy Pechnikov

$ git log --career --oneline --stat

Work Experience

a3f7c1e Oct 2024 - Sep 2025 Lead Backend Engineer (Contract)

Fenixwb — Wildberries slot booking service - 200 jobs/sec, 10K+ users · Remote

Inherited an unstable monolith prototype pushing ~10 tasks/sec - redesigned the entire backend on Temporal.io with ~10 microservices integrated with Elasticsearch and MySQL
Pushed throughput from 10 to 200 background jobs/sec (20x), the threshold that let the product handle real user load and start onboarding paying customers
Disassembled a WASM captcha binary in Ghidra, then rewrote the solver in Rust - execution dropped from 2s to 0.1s (20x faster), giving clients higher booking success rates than competing bots
Migrated from MongoDB to PostgreSQL - simpler queries, fewer ops headaches for a 3-person backend team
Sourced and hired 2 engineers (backend + DevOps) from scratch, building the engineering function from zero - owned architecture calls, task allocation, and code review
Traced a critical throughput bottleneck via hypothesis-driven load testing - the root cause fix is what took us from ~10 to 200 tasks/sec
Went from zero monitoring to production-grade observability: Grafana + Victoria Metrics + Loki with custom dashboards and alerting - incidents detected in ~3 minutes vs previously hours, MTTR dropped from days to hours
Chose to scrap the failing monolith for Temporal.io over incremental patches - the architectural call that enabled 10→200 tasks/sec across ~10 services

RustPythonTemporal.ioElasticsearchMySQLPostgreSQLasyncioGrafanaVictoria MetricsLokiDockerKubernetes

b8e2d4a May 2023 - Nov 2024 Founding CTO / Technical Lead

Cryptosofters — Multi-wallet orchestration platform with visual pipeline builder - near-production, drag-and-drop wallet operations · Remote (Dubai-based)

Designed multi-wallet orchestration platform with drag-and-drop pipeline constructor - users configured chains of on-chain operations (swaps, bridges, contract calls) without code. Built to near-production before investment stopped
Under the hood: event-driven state machine with per-wallet operation queues, automatic retry with exponential backoff, and transaction rollback on partial failures - the engine that made drag-and-drop orchestration reliable enough for real money
Zero security incidents across the platform's lifetime. Architected per-user wallet isolation, credential management, and transaction signing in a multi-tenant environment where investor confidence hinged on it
Reverse-engineered closed-source smart contracts via on-chain transaction analysis and ABI decoding - often the only way to integrate protocols with no SDK, no docs, and no public API
Architected the SDK generation pipeline: mitmproxy traffic capture → OpenAPI spec → auto-generated Python SDK for 20+ CEX APIs. Cut exchange integration from ~1 month to ~1 week (75%) - the force multiplier that made a 9-person team viable on a $100K budget
Delivered 20+ DeFi protocol integrations (Uniswap, 1Inch, Stargate, LayerZero) with a unified interface via web3.py - per-protocol time dropped from ~1 week to ~1 day (~80%), and any team member could add protocols, not just the specialists
9 engineers (4 backend, 3 frontend, 1 Solidity, 1 PM), $100K annual budget, zero turnover - kept the team together through a budget crisis that would've scattered most startups
Grew the team from 0 to 9: wrote job descriptions, ran technical interviews, onboarded everyone, and shaped engineering culture from day one
Set up code review standards, CI/CD pipelines, sprint cadence, and knowledge-sharing sessions - bug escape rate dropped and delivery became predictable enough to commit to external timelines
Runway hit zero. Assembled an investor pitch deck, led the fundraising conversations myself, and bought the team more time - the moment I learned operational discipline matters as much as architecture
Introduced code review process and CI/CD pipelines for a team of 9 - regression bugs dropped noticeably and new engineers onboarded faster

PythonWeb3.pymitmproxyOpenAPIasyncioPostgreSQLRedisDocker

c5a1f9b Nov 2021 - Apr 2023 Python Developer / Tech Lead

Vitrinagram — Telegram MiniApp e-commerce platform - multi-tenant storefronts, 1,000+ concurrent users per store · Remote (Moscow-based)

Built multi-tenant Telegram MiniApp backends handling 1,000+ concurrent users per store - horizontal scaling via Celery + Kubernetes meant cost grew linearly with tenants, not exponentially
Automated tenant provisioning: FluxCD + Cloudflare API pipeline, ~3 minutes from commit to live subdomain
Developed an S3-compatible proxy API on MinIO for tenant-isolated file storage - eliminated the shared-storage security risk and cut costs vs managed S3
Per-tenant observability out of the box: Grafana + Victoria Metrics + Loki + Sentry dashboards clients could use for self-service debugging - ~40% fewer support tickets once clients could self-debug
Led backend development across a team of 7, owning architecture decisions for multi-tenant scalability, setting technical direction for the platform, and mentoring mid-level engineers through code review and design discussions that improved team velocity over two quarters
Spotted a technical leadership vacuum that was slowing delivery - escalated to the CEO, which led to team restructuring and a noticeable improvement in sprint completion rates

PythonFastAPICeleryPostgreSQLRedisDockerKubernetesFluxCDGrafanaVictoria MetricsLokiSentryMinIOCloudflare

d2c8e7f Jan 2018 - Present Backend Developer / Lead

Freelance — Crypto infrastructure, trading systems, and data pipelines

Built EVM wallet management system for a client with multiple investors: 2,000+ wallets, 5,000+ tx/day, fully autonomous - equivalent of eliminating a full-time ops person
Reverse-engineered TradingView's WebSocket protocol and deobfuscated their JavaScript - collected 260M candlesticks in ~1 week covering 20+ years of forex and crypto data. Comparable feeds charge $600+/month
Led 3 Go developers building a CEX arbitrage bot with sub-100ms execution latency - cross-language coordination (Python orchestration, Go hot path) where every millisecond was the difference between profit and a missed window
Portfolio managers needed consistent exposure across 10+ exchange accounts - built an automated replication system mirroring positions in real-time with near-zero drift between source and mirror
Developed a USDT-TON payment gateway with near-instant confirmation for a gaming platform - settlements in seconds vs traditional processing delays (delivered, pre-launch)
Chose Rust for the Polymarket transaction parser because Python couldn't sustain the throughput - 300 blk/s from distributed event sources, production data ingestion
Deployed single-node k3s on NixOS with declarative config, Terraform provisioning, and GitOps workflow - reproducible infra, entire cluster from zero in under an hour
Provisioned Kubernetes clusters on Yandex Cloud and Hetzner using Terraform for multiple clients - standardized the infra setup, cut provisioning time across engagements
Building quantitative backtesting infrastructure on nautilus-trader with 260M+ candles stored in TimescaleDB - time-series queries that would choke PostgreSQL run in seconds
100% completion rate across 8+ freelance engagements. Repeat clients. Handled everything end-to-end - from initial scoping call to production handoff

PythonRustJS/TSWeb3.pyethers.jsPostgreSQLRedisasyncioNixOSk3sTerraformTimescaleDBDocker

Work Experience

Technical Skills

Languages

Frameworks & Libraries

Databases

Cloud & DevOps

Observability

Blockchain & DeFi

Security & Analysis

Featured Projects

Professional Development

Let's Connect