Dilshat Rakhimov

The agency model was burning budget on brand queries with no visibility into actual user quality. The app had severe technical issues — users frequently reinstalled, each time facing full KYC with video verification. Registration had dozens of unnecessary steps and platform-specific bugs. There was no attribution system linking ad spend to product engagement or revenue.

• Built attribution from scratch: Negotiated a direct contract with Adjust (secured free access to audiences, smart banners, and smart links by leveraging competitive pressure from Appsflyer). Configured attribution across Google UAC, Meta, TikTok, Yandex, and affiliate channels. Used Firebase SDK for Google conversions, Adjust SDK for Amplitude reporting.
• Designed event taxonomy for 15+ product lines: Loans, deposits, investments, payments, transfers, gov-tech, insurance, cinema tickets, travel — for both B2C and B2B segments. Reviewed data governance monthly to prevent taxonomy degradation.
• Eliminated agency dependency: Transitioned from 30% agency commission to in-house team with 1% billing. Grew the annual UA budget from ~$200K to $3M+ by proving ROI to leadership. Secured direct contracts with Google, Yandex, and TikTok.
• Fixed onboarding funnel: Removed unnecessary warning steps (+5% conversion). Integrated Kafka-stored user data to skip KYC for existing bank clients — this single change drove +30% conversion lift. Presented lost-user cost analysis to CEO to secure engineering resources.
• Solved re-verification problem: Championed automated identity verification to eliminate repeated video KYC on reinstalls, reducing friction for millions of returning users.
• Drove 30-35% of bank's total sales through digital: Managed end-to-end P&L for retail and business products. 30% of loans disbursed from digital campaigns.

Web-view attribution remained broken — the bank's ecosystem services used embedded web-views that broke session continuity. I authored complete technical documentation (JS bridge integration, cross-authorization handling, UTM passthrough architecture) but developers never implemented the solution.

• GTM + Pixel infrastructure: Implemented Google Tag Manager from scratch. Configured all ad pixels (Google, Meta, TikTok), managed Content Security Policy, and built dataLayer events triggered only on confirmed CRM API success — ensuring ad platforms trained on real conversions, not form submissions.
• Attribution pipeline → DWH → Dashboards: Connected Google Ads API, Meta API, TikTok API, and Appsflyer raw data into the bank's data lake. Authored technical documentation; analytics team built connectors. Built automated PowerBI and Qlik dashboards linking marketing spend to downstream revenue.
• Smart lead routing with government APIs: Conceptualized and co-architected a verification engine integrating Kazakhstan's tax database (Salyk) and citizen registry. Wrote GTM-side conversion scripts. The system auto-validated whether a user was an individual entrepreneur or legal entity, routing them to correct funnels and ensuring ad campaigns trained on proper audience signals.
• CMS migration for SEO: Convinced engineering to migrate from micro-service SPA (every content change required dev team + QA + full release cycle) to October CMS, dramatically reducing time-to-market for landing pages.
• Campaign optimization training: Educated the in-house team on value-based bidding with Firebase events — a shift from generic install optimization that immediately improved ad delivery and conversion quality.

Built a Python pipeline that connects to Figma API, extracts key frames from design files, analyzes screen content (forms, CTAs, navigation, error states), and auto-generates a structured event taxonomy for Amplitude and GA4 — complete with event names, properties, and funnel definitions. Output is an Excel tracking plan ready for developer implementation.

• Figma REST API integration for frame extraction and text analysis
• Rule-based event detection engine (forms, payments, navigation, errors, uploads)
• Automatic property mapping with PII filtering
• Consistent naming convention enforcement across all flows
• Excel export with per-flow event sheets

• Fetch API interception via GTM: Built a Custom HTML tag that monkey-patches window.fetch to intercept specific API endpoints (calculator, schedule preview, collateral check, application creation) and push structured dataLayer events only on successful HTTP responses.
• Value-based conversion tracking: Captured loan amount from the create-credit payload and passed it as conversion value (KZT) to Google Ads, Meta, TikTok, and Yandex — enabling value-based bidding optimization.
• Multi-platform pixel setup: Single GTM implementation feeding GA4, Google Ads, Meta Conversions API, TikTok Events API, and Yandex Metrika simultaneously.
• PII protection: Deliberately excluded personal identifiers (borrower IDs, document numbers) from all analytics events while preserving business-critical parameters.

A full digital transformation strategy for a national bank, covering six interconnected workstreams designed to close the gap between ad spend and actual revenue:

• S2S Offline Conversion Pipeline: Architecture for feeding CRM loan statuses (created → qualified → approved → disbursed) back to Google Ads, Meta CAPI, and TikTok Events API. Enables value-based bidding on actual loan disbursement amounts instead of form submissions. Includes Python code for API integration and data flow diagrams.
• Pre-Scoring & Web-to-App Onboarding: Anti-fraud module integrating government identity databases for real-time applicant verification. Alternative data scoring via bank statement parsing (PDF). Seamless handoff from web to mobile app via AppsFlyer deferred deep links, QR codes, and SMS with pre-approved amount.
• Predicted Lifetime Value (pLTV): Framework for predicting customer LTV within 24 hours of install using micro-conversion signals (biometric completion, credit terms viewed, first application screen). Integration with AppsFlyer S2S API and Google Ads for algorithmic optimization on high-value users.
• End-to-End Attribution with MMP + DWH: Full pipeline connecting AppsFlyer Data Locker to the bank's enterprise DWH via Airflow ETL. Revenue events from the core banking system flow back to AppsFlyer via S2S API, enabling true ROAS measurement across all channels.
• Unified User Identity in Amplitude: Migration plan from fragmented analytics (multiple Amplitude projects, PII as user IDs, broken cross-service sessions) to a single-project architecture with synthetic UUIDs, shared user context across app modules, and historical data merge via Amplitude Alias API.
• Native-to-WebView Session Continuity: Technical spec for passing Amplitude device IDs, user IDs, and attribution data from Flutter native app into embedded web-views via JavaScript bridge injection. Includes working Flutter and JS code samples.

Production-ready architecture for attributing offline branch visits to online ad clicks across Google, Meta, and TikTok. The most comprehensive document in this collection — includes everything from JavaScript click trackers to Airflow DAGs to platform upload scripts.

• Click Tracking Layer: JavaScript tracker deployed on website and in-app web-views. Captures gclid/fbclid/ttclid + Meta cookies (_fbc/_fbp) + TikTok cookies (_ttp). Multi-layer persistence: localStorage → cookies → IndexedDB → Service Worker cache. Cross-subdomain cookie support.
• Device Fingerprinting: Canvas + WebGL + hardware fingerprinting as secondary matching signal for users who clear cookies between online click and offline visit. Probabilistic matching when deterministic IDs are unavailable.
• Database Schema: Complete PostgreSQL/Redshift schema — ad_clicks, user_click_mapping, offline_events, attributed_conversions, retargeting_audiences tables with proper indexing strategy.
• Attribution Engine: Nightly Airflow batch job running last-click attribution SQL with 7-day lookback window. Match hierarchy: user ID match → device ID match → fingerprint match, with confidence scoring.
• Platform Upload: Python scripts for uploading attributed conversions back to Google Ads Offline Conversion API, Meta Conversions API, and TikTok Events API. Includes retry logic, partial failure handling, and upload status tracking.
• Retargeting Audiences: Three automated audience segments — "visited branch, no purchase", "high-value customers", "clicked ad, never visited" — synced to Google Customer Match, Meta Custom Audiences, and TikTok via hashed PII.

Step-by-step technical specification for integrating a branded domain with AppsFlyer OneLink in a Flutter mobile app. Designed as a developer handoff document with zero ambiguity — every config change, code snippet, and test case is specified.

• iOS: Associated Domains configuration in Xcode, Apple Developer Portal setup, AASA file verification
• Android: AndroidManifest.xml intent-filters with autoVerify, SHA256 fingerprint verification, assetlinks.json validation
• Flutter SDK: setOneLinkCustomDomain() initialization order (must be called before initSdk()), Unified Deep Linking callback handling, URI Scheme fallback configuration
• QA Test Plan: 15+ test cases across iOS Universal Links, Android App Links, SDK conversion data, deferred deep links, and cross-browser testing matrix (Safari, Chrome, Telegram, WhatsApp, Instagram in-app browser, SMS, QR codes)

Two technical specifications for real-time lead validation using national business registry APIs. The system determines whether a user is a registered business entity or an individual consumer at the point of form submission, then routes them accordingly — businesses proceed to the sales funnel, individuals get redirected to the retail mobile app via AppsFlyer smart links.

• Client-side validation: Phone number regex for regional mobile formats, input masking, real-time UX feedback
• Government API integration: Backend proxy for national statistics bureau API with response caching (24h TTL), rate limiting, and error handling for API downtime
• Smart routing logic: Business entities → form submission + conversion event. Individuals → modal popup with QR code (desktop) or direct deep link (mobile) to retail banking app via AppsFlyer OneLink with full attribution parameters
• DataLayer integration: Conditional conversion events — only fires on qualified leads, preventing ad platforms from optimizing on unqualified traffic

Technical task specification for resolving critical Core Web Vitals issues on a banking website. Organized as a prioritized developer handoff with specific Nginx configs, HTML/CSS code, and measurable acceptance criteria for each task.

• TTFB optimization: Server-side HTML caching (Nginx + CDN), Redis object caching with per-content-type TTL, backend monitoring setup (2,700ms → target ≤500ms)
• LCP optimization: Responsive images with AVIF/WebP srcset, Brotli compression, preload for LCP element, critical CSS extraction (4,900ms → target ≤2,500ms)
• JS/CSS optimization: Deferred GTM and analytics loading via requestIdleCallback, tree-shaking, code splitting, dynamic imports for widgets
• FAQ structured data: Dynamic URL routing for SEO, JSON-LD FAQPage schema, SSR requirements for crawler indexability

A GTM Custom HTML implementation that monkey-patches window.fetch to intercept SPA API calls and generate dataLayer events for a 4-step secured loan application funnel. Fires conversion events with loan amount as value only on successful HTTP responses.

• Fetch interception for calculator, schedule preview, collateral check, and application creation endpoints
• Value-based conversion: captures loan amount from API payload, passes as conversion value to Google Ads, Meta, TikTok, Yandex simultaneously
• PII filtering: deliberately excludes borrower IDs and document numbers from all analytics
• Complete GTM setup guide: DataLayer Variables, Custom Event triggers, GA4 tags, and multi-platform pixel configuration

Engagement model: I work as a fractional consultant on retainer (20+ hrs/week available) or project-based. I'm based in Almaty, Kazakhstan (GMT+5) with significant overlap with European business hours and flexible for US calls.

What you get: Not just strategy decks — I write production code (Python, JavaScript), build GTM implementations, author technical specs for your dev team, configure attribution platforms, and create dashboards. I bridge the gap between marketing strategy and engineering execution.

Industries: Deep expertise in fintech (banking, lending, insurance, payments) and marketplace/e-commerce apps. Particularly strong where mobile attribution, privacy constraints (iOS 14.5+), and online-to-offline conversion tracking are critical challenges.

Languages: English (fluent, professional), Russian (native), Japanese (conversational, JLPT N2).