Jamoliddin Karimov

AI Systems Engineer I design AI-powered systems that automate marketing, engagement, and operational workflows with deterministic control and validation layers. My work focuses on building stateful AI agents, multimodal processing pipelines, and scalable data infrastructure using APIs, orchestration tools, and structured generation frameworks. 1. Meta Business Suite Message Automation & TTFR System Full technical breakdown: Meta Business Suite Automation Detailed One-Line Summary Built a real-time monitoring and response-tracking system for multiple Meta (Facebook & Instagram) assets, automating inbound notifications and measuring Time to First Response (TTFR) with AI-based message filtering. The Problem The client needed visibility into inbound messages across multiple Meta Business assets. Challenges included: • No centralized notification system • No structured tracking of response times • Spam and low-value messages creating noise • No measurable way to monitor support responsiveness There was no reliable system to track how quickly teams responded to real customer inquiries. What I Built A real-time automation system that: • Subscribes to Meta webhooks (Facebook & Instagram) • Detects new inbound messages across multiple assets • Uses AI to classify messages as valid or spam • Sends structured email notifications for valid messages • Tracks first manual reply timestamps • Calculates TTFR (Time to First Response) • Logs all activity into daily Google Sheets reports • Sends automated daily TTFR summary emails How It Works (High-Level) 1. Webhooks receive inbound messages from Meta. 2. AI filters out spam and non-actionable messages. 3. Valid messages trigger email notifications. 4. First reply timestamps are recorded. 5. TTFR is calculated automatically. 6. A daily summary report is generated and emailed. The system supports multiple portfolios and assets with scalable token management. Why It Matters • Improved response monitoring across multiple social accounts • Reduced noise through AI-based filtering • Introduced measurable accountability (TTFR tracking) • Automated daily reporting • Enabled operational visibility across support channels Shifted from reactive message handling to structured performance monitoring. Tech Stack n8n Meta Graph API (Webhooks + Conversations endpoints) OpenAI API (message classification) Google Sheets API Gmail API 2. Klaviyo Email Intelligence System Full technical breakdown: Klaviyo detailed One-Line Summary Built an automated system that connects Klaviyo performance data with email creative structure and generates AI-driven insights — replacing a manual export + GPT workflow. The Problem Email reporting was manual and fragmented: • Metrics exported from Klaviyo • Creative reviewed separately • GPT used manually for analysis • No scalable structure Performance data and creative content were disconnected. What I Built A modular automation system that: • Pulls campaign and flow data via API • Extracts email metadata and template content • Converts creative structure into measurable attributes • Aggregates structured data • Feeds it into an AI analysis layer • Logs results into Google Sheets Supports campaign mode, flow mode, and comparison mode. Why It Matters • Eliminated manual exports • Standardized analysis process • Linked creative decisions to performance metrics • Reduced reporting to a single execution Shifted from reporting automation to a reusable email intelligence system. Tech Stack n8n Klaviyo API OpenAI API Google Sheets 3. Automated Monthly Keyword Rank Tracking System Full technical breakdown: Automated Monthly Keyword Rank Tracking System (SEMrush + Google Sheets) One-Line Summary Built a scalable monthly SEO rank tracking system that automatically retrieves keyword positions from SEMrush and appends structured ranking history into Google Sheets without overwriting past data. The Problem Keyword rankings were being checked and logged manually each month. This created: • Repetitive manual work • Inconsistent historical tracking • Complexity across devices and domains • Risk of overwriting or misreporting data There was no reliable infrastructure for long-term SEO monitoring. What I Built A modular automation system that: • Pulls keyword rankings via SEMrush API • Tracks Desktop and Mobile performance • Supports multi-domain setups (Retail + Wholesale variant) • Dynamically creates new “Month X” and “Ranking URL” columns • Preserves full historical ranking data • Allows SKIP logic for selected keywords • Automatically calculates average ranking values Why It Matters • Eliminated manual monthly rank checks • Created structured month-over-month SEO visibility • Preserved historical performance data • Reduced reporting errors • Scalable to additional domains and keyword sets Tech Stack n8n SEMrush API Google Sheets 4. Meta Video Ads Analyzer Full technical breakdown: Meta Video Ads Analyzer One-Line Summary Built an AI-powered competitive intelligence system that scrapes Meta video ads, analyzes their creative structure, and stores structured insights for strategic benchmarking. The Problem Marketing teams manually reviewed competitor ads in Meta Ad Library to understand creative trends. This process was: • Time-consuming • Unstructured • Difficult to compare across competitors • Lacking systematic documentation There was no scalable way to analyze what made competitor video ads effective. What I Built An automated workflow that: • Scrapes Meta Ad Library video ads via Apify • Filters and ranks ads by reach or runtime • Downloads and processes video files • Uses AI (Google Gemini) to extract structured creative insights • Identifies hook, transcript, ad format, concept, and narrative structure • Stores structured outputs in Google Sheets • Generates a dedicated analysis sheet per competitor Why It Matters • Eliminated manual competitor ad review • Standardized creative analysis across multiple brands • Enabled systematic benchmarking of ad formats and hooks • Turned unstructured video content into structured strategic data Tech Stack n8n Meta Ad Library (via Apify) Google Gemini API Gemini API OpenAI (structured parsing) Google Drive Google Sheets 5. Avery AI Agent for X Engagement Full technical breakdown: Avery AI Agent for X Engagement One-Line Summary Built a stateful AI engagement agent on X (Twitter) that autonomously monitors conversations, analyzes text and images, and generates context-aware replies while managing thread depth and platform compliance. The Problem Manual engagement on X was not scalable. The brand needed to: • Respond to mentions quickly • Join conversations when tagged • Engage with influencer accounts • Maintain consistent tone and branding • Avoid spam behavior or duplicate replies There was no structured system to manage conversational context at scale. What I Built A multi-branch AI orchestration system that: • Monitors mentions, replies, tags, and curated lists • Processes tweet text and images via Grok (multimodal analysis) • Generates persona-controlled replies using OpenAI • Tracks processed tweet IDs to prevent duplication • Maintains reply count per thread • Limits engagement depth automatically • Generates autonomous original tweets on schedule • Logs activity for traceability Why It Matters • Created a scalable AI-driven social engagement engine • Enabled real-time, context-aware replies • Maintained brand voice consistency • Prevented spam loops and over-engagement • Combined reactive and proactive content automation Tech Stack n8n X API v2 xAI Grok API OpenAI API Google Sheets 6. Email Performance & Creative Intelligence Agent Full technical breakdown: Email Performance & Creative Intelligence Agent One-Line Summary Built an AI-powered email calendar and content generation system that automatically distributes campaigns and flows across a full month, aligns with holidays, enforces strict output validation, and publishes structured results directly to Google Sheets. The Problem Email marketing calendars were being created manually using custom GPT prompts. This caused: • Inconsistent scheduling logic • Manual tone analysis per client • Risk of incorrect campaign/flow distribution • Errors in formatting before uploading to Sheets • No strict validation of AI output The process was time-consuming and difficult to scale across multiple clients. What I Built A structured AI orchestration workflow that: • Collects client inputs via form (email count, split, website, promotions) • Automatically analyzes the client website to extract brand tone guidelines • Programmatically calculates evenly distributed email dates • Generates a full-month calendar framework (all days included) • Uses an AI agent to generate email copy under strict constraints • Enforces exact email counts and campaign/flow distribution • Validates AI output against calendar rules • Merges generated content into a complete month calendar • Automatically creates or updates the client’s Google Sheet • Sends success/failure notifications Why It Matters • Eliminated manual GPT-based planning • Introduced deterministic scheduling logic • Prevented AI over-generation or date drift • Standardized email calendar structure across clients • Made monthly execution repeatable and scalable Tech Stack n8n OpenAI API Google Sheets Google Drive Custom JavaScript validation & scheduling logic