Kostandin Hidri · New York, NY
Technical Account Manager for Global Strategic Accounts at Smartly.io. Before that: enterprise onboarding at scale, support engineering, and a sneaker bot that taught me everything about automation, products, and the people who depend on them.
2017 – 2021 · Co-Owner & Developer
A Nike automation bot born from a simple observation: the most coveted sneakers on earth sell out in under sixty seconds, and no human was ever going to win that race by hand.
In 2017 the sneaker resale market was turning into an asset class. Limited Jordan and collab releases were trading at three to ten times retail on StockX and GOAT, and every drop was a race measured in milliseconds. Watching pairs vanish before a page could even load, I realized this wasn't a shopping problem; it was an automation problem. Whoever could detect stock, fill a cart, and check out faster than a browser refresh would win, every time.
We built a cross-platform desktop app in Node.js + Electron: a task engine that monitored releases, managed dozens of profiles and proxies in parallel, and automated the full checkout flow against Nike.com. A Bootstrap UI made it usable by people who'd never opened a terminal: spin up twenty tasks, walk away, get a Discord ping when the order confirms.
When everyone is automated, the race moves down the stack. I negotiated directly with data-center suppliers to secure servers with best-in-class connection speeds to Nike.com, turning network latency itself into our competitive moat. Milliseconds were the product.
The real transformation was commercial: we productized the bot into a SaaS offering with a license-key model (356 keys sold), launched it publicly across social platforms, and ran the whole customer lifecycle out of Discord: onboarding, tutorials, support, retention. Build the automation, package it, then take care of the people who depend on it. That loop became the blueprint for everything I've built since.
$ sneakerware run --drop "AJ1 Retro High OG"
[09:59:58.412] 24 tasks armed · 24 proxies healthy
[10:00:00.087] drop live · monitoring stock
[10:00:00.291] size 9.5 carted · task 07
[10:00:00.304] size 10 carted · task 12
[10:00:01.118] checkout submitted ×6
[10:00:02.450] ✓ order confirmed · #NK4821
[10:00:02.731] ✓ order confirmed · #NK4834
[10:00:03.006] 6/24 success · webhook → Discord
█
Smartly.io · n8n · Slack · AI agents
An AI-powered operations pipeline that automates the journey from Salesforce closed-won to onboarding wrap-up (intake, triage, enrichment, and routing) so humans only touch the judgment calls.
The real lifecycle: Salesforce closed-won to CSM handoff, mapped across Standard, Scale, and Agency tracks. Amber bolts mark steps the n8n pipeline runs without a human.
I owned onboarding for 30+ enterprise customers (Rocket Mortgage, DraftKings, The New York Times, Roblox), each with 50+ users and roughly $630M in combined managed ad spend. Before automating anything, I mapped the whole machine: every step from Salesforce closed-won to CSM handoff, across three delivery tracks (Standard, Scale self-guided, and Agency), with each step tagged by its owner. Then I added a fifth tag, "things to automate," and it lit up everywhere. Welcome emails, onboarding requests, closed-won handoffs, client notifications: most of the process needed coordination, not judgment.
I made n8n the orchestration layer and Slack the interface. A Salesforce closed-won fires the pipeline: the welcome email and onboarding request generate themselves, AI agents parse and triage intake, enrich it with account and product context, route work to the right owner, and nudge anything that goes stale, all the way through account setup, trainings, testing, and wrap-up. People stopped pushing tickets and started making decisions.
Manual support touchpoints dropped by ~40%. Average time-to-close for net-new onboardings fell by ~51%. Churn across the book stayed at zero, and onboarding success contributed ~$3.5M in revenue through expansion and proof-of-concept delivery. The same playbook now informs how I run global strategic accounts today.
Full-stack · Next.js 15 · Claude AI · Google Sheets
A web app that turns raw creative into schema-validated, multi-platform ad trafficking feeds for Meta, TikTok, Snapchat, and Reddit, written straight to Google Sheets, with Claude as a guardrailed copilot.
At Smartly I watched the world's largest advertisers run launches through hand-built spreadsheet feeds. The schemas are strict: one wrong column name and the automation downstream breaks silently. But the schema is known, and every asset already carries its own metadata. So the entire feed could be generated and validated by software. Humans should review feeds, not type them.
A full product, not a script: Next.js 15 + React 19 + TypeScript, multi-tenant company workspaces on Supabase, Google OAuth, and deep Sheets + Drive integration. An 8-step trafficking wizard detects aspect ratios, merges creative variants into one row per creative, groups carousels with per-platform limits, and auto-calculates flight status, then writes 30-column, header-validated feeds to the team's sheet.
Claude powers copy generation and vision-based carousel detection, clustering assets and proposing a storytelling order from hook to CTA. But it works inside hard constraints: every AI call uses tool_use (function calling) with schema-enforced output, so the model literally cannot invent a column or exceed a character cap. Schema-first, validation-enforced, append-safe.
Inside Smartly, feeds are how campaigns get orchestrated across every platform: clean feeds mean faster launches and fewer silent failures, and patterns from this build went on to inform core product capabilities adopted by broader customer segments. Outside Smartly, it's platform-agnostic by design: Sheets remain the source of truth, so any agency or in-house team running spreadsheet-based trafficking can use it, and the schema registry makes a new platform a config change, not a rebuild.
The path
Owning the full technical lifecycle for global strategic accounts like Procter & Gamble and agencies like WPromote: S2S integrations, conversions API pipelines, reusable solutions deployed across the account base, and in-person technical workshops from C-suite to engineering.
Owned onboarding for 30+ enterprise customers across ~$630M in managed spend: Rocket Mortgage, DraftKings, FanDuel, The New York Times, Roblox, Bloomberg. Built the AI-powered onboarding pipeline and the feed automation that informed core product.
Tier 1–3 enterprise incident resolution end-to-end. 98.8% CSAT. Grafana, Kibana, and SQL to find systemic issues, then automated them away inside Intercom.
SQL, EDI, and connectivity support across API, AS2, sFTP, and VAN: the enterprise integration trenches.
Where it all started: a Nike automation bot turned SaaS. Product, growth, support, and retention, learned by doing all of it.
The exact combination this career runs on.