# Mustafa DOGAN — machine-readable CV > Canonical: https://mddogan.com/cv > Graph: https://mddogan.com/api/graph > JSON Resume:https://mddogan.com/api/graph?format=jsonresume > Markdown: https://mddogan.com/cv.md > MCP: https://mddogan.com/api/mcp > Updated: 2026-08-11 ## Role Technical Account Manager — Technical Account Manager · CCaaS platforms · API and system integration Based in Metro Manila, Philippines. ## Now Technical Account Manager, Call Center Studio (Aug 2025 - present). Technical bridge between internal teams and external stakeholders, and the escalation point between clients, product and engineering on complex implementation work. Hands-on delivery for clients in the Philippines. I design and build AI-powered internal and client-facing applications, integrate APIs and SQL systems, and build AI-orchestrated workflows with n8n and MCP/WebMCP. ## Education - MSc Machine Learning & Data Science, Imperial College London (Merit, 2025) - BSc Management Information Systems, Karadeniz Technical University (Honours, 2023) ## Capabilities - Client technical ownership: I am the technical point of contact for client stakeholders and the escalation route into product and engineering. That covers scoping what a tenant needs, turning it into something engineering can act on, and staying on it through delivery and adoption. Four years on the same platform across support, business analysis and account management means I can work a problem from the ticket detail up to the stakeholder conversation. Enablement is part of it rather than a handover: I have run a two-day enablement and certification programme for an enterprise customer's contact-centre organisation, around 130 people, provisioning their accounts in bulk before each session. I travel to a customer's site when a multi-vendor cutover needs someone in the room. - Platform automation and bulk operations: Around 20 projects do administration through a platform's own interfaces instead of its console: bulk configuration rollouts, roster and extension assignment, scheduled reconciliation. 140 n8n workflows across development and production, 62 of them running, 929 nodes in total with 272 custom code nodes and 106 webhook endpoints that other systems call. Every bulk action ships with its rails: preview, drift detection, run caps, audit rows and a rollback path. - Integration middleware: Nine projects sit between two systems that have no integration path: a CRM and a contact-centre platform, a warehouse and an operations console. The recurring shape is a broker that owns OAuth install and refresh, per-tenant encrypted credentials, and a field-mapping layer, structured behind provider interfaces so a second CRM or a second platform arrives as an adapter rather than a rewrite. The same period covers CRM and ERP administration as a business analyst, across HubSpot, Zoho and SAP. - Recovering an undocumented interface: 14 captured traffic sessions sit in this portfolio, each one the start of the same method: record the live traffic, check it against whatever specification exists, write down where the two diverge, and encode the verified contract as a module the dependent code cites. Where a system accepts unknown parameters silently, a parameter counts as supported only once a deliberately non-matching value is shown to shrink the result count, so the map is probed rather than assumed. Each of these paths sits behind a provider interface, so a sanctioned API can replace it later without touching the callers. - Mobile delivery: Two Flutter applications shipped for phones, both built around an existing web product rather than replacing it. One is a full client for the reporting product on iOS and Android, matching the web app screen for screen so there is one set of behaviour to support instead of two. The other hosts the browser softphone on Android so an agent can work from a phone, keeping the call session alive in the background and wiring the microphone and notifications through to the native side. - Reporting and reconciliation: 11 projects do the same job at different scales: take a number somebody maintains by hand and make it trustworthy. Roster reconciliation between systems, order-accuracy verification against a live pricing service, operational reporting over a warehouse. Where a model helps, the deterministic layer keeps control: a generated query does not execute until it passes a cost estimate against configured limits, and safety decisions come from rule-based gates. - Test automation and load testing: Browser tests for an ordering integration are stored and edited centrally rather than checked in as scripts, run in real browsers on demand, watched while they run, and their artefacts filed per run; access is by role with an approval step before a new account can run anything. Captured requests have also been replayed as load, so endpoints were measured under repetition rather than proven to work once, with the results kept alongside the capture they came from. - Working across timezones: I am based in Manila and the engineering team is in Türkiye, so the working day is built around an overlap rather than an office. Over one recent six-week stretch my calendar carried 10 distinct scheduling timezones, and more meetings were set in Istanbul time than in my own — 121 against 99. Just over a third of them started at 18:00 or later where I sit. The counterparts in that period were in the Philippines, Türkiye, Pakistan, the United Kingdom, the United States, Norway, Germany, Malaysia, India, Taiwan, Brunei and China. Internal sessions run in Turkish and customer sessions in English, often in the same week. - Applications and browser tooling: 27 applications in production across 416 deployments, from 49 private repositories: TypeScript and Next.js on the web, Flutter and Kotlin for mobile, Python for data work. Eight Manifest V3 Chrome extensions, one published to the Chrome Web Store, including a network-capture developer tool and typed tool bridges built on the Model Context Protocol (MCP and WebMCP) that let a model drive a web application with no automation API. Retrieval-augmented generation is used here to ground answers in a known corpus. ## Credentials 158 professional certifications. 158 certifications are listed on LinkedIn. The entries below are the most recent, taken from the profile with their credential IDs where LinkedIn shows them; the rest remain on the LinkedIn profile rather than being duplicated here. ## Projects - Dialler rule engine — Query compilation - Agentic Report Studio — Second place, AI hackathon - Reporting client for iOS and Android — Mobile delivery - Curyy — A product owned end to end - CRM to contact-centre bridge — Integration middleware - May reklamo — A product shipped end to end - Call search and bulk recording export — Multi-tenant product delivery - Softphone shell for Android — Native and web integration - Demo builder — Pre-sales tooling - Browser test platform — Test automation - API Explorer — Reverse engineering, as a tool - Workflow-as-code toolkit — Infrastructure as code - On-call schedule and alerts — Operations tooling - Call flow documentation and build automation — Configuration as code - Model Context Tool Inspector — Agent tooling at the edge of the platform - Self-hosted automation platform — Workflow automation - RAG Lab — Explaining a system by building it - Digital business card — Front-end delivery ## Summaries by role Each role below is answered in three parts: what the record covers, what is not on it yet with the nearest thing that is, and the gaps. The gaps are not omitted for any role. - Technical Account Manager (/d/technical-account-manager) — 5 covered, 2 learnable, 3 gaps - Customer Success Lead (/d/customer-success-lead) — 5 covered, 3 learnable, 3 gaps - Automation and Platform Lead (/d/automation-lead) — 5 covered, 3 learnable, 4 gaps - Technical Delivery Manager (/d/delivery-manager) — 5 covered, 3 learnable, 4 gaps - Product Manager, Platform and Integrations (/d/product-manager) — 5 covered, 3 learnable, 4 gaps - Partner Enablement Manager (/d/partner-enablement) — 5 covered, 3 learnable, 3 gaps ## Languages Turkish (Native), English (Professional), Kurdish (Professional), Spanish (Elementary), Tagalog (Elementary) ## Contact mail@mdogan.com ## For agents All crawlers are welcome; nothing here is gated. Tools available over MCP: list_surfaces, get_graph, search_cv, get_experience_by_skill, get_certifications, compare_to_job_description, mint_dossier, get_project, download_cv. Human and agent surfaces are held at parity — anything a person can read or do here is reachable over the protocol, and that parity is asserted in CI.