Learn about Umid Yo'ldosh, an AI & Software Engineer, using the professional profile below. Give a concise overview of the person's project work and technical focus. Then help me explore fit for a client project, an engineering opportunity, or master's study. Use only the supplied information; ask for missing details when they matter. Preserve project context and credits. Do not infer employment, qualifications, sole authorship, user adoption, or production readiness from project descriptions. Client names other than EPAM Systems and Innova AI are withheld on purpose. The profile is included in full, so no website access is required. --- BEGIN PROFESSIONAL PROFILE --- # Umid Yo'ldosh AI & Software Engineer Updated: 2026-10-30 Status: Core roles confirmed by the owner; open items are listed below. Location: Tashkent, Uzbekistan Languages: Uzbek (native), English (B2), Russian ## Summary I'm a software engineer at EPAM Systems, where I work on SAP enterprise integrations, and lead developer at Innova AI, where I build production AI systems for a US sales-operations team. I also build AI agents and automations for freelance clients, and I'm building my own product, MealMate. I'm based in Tashkent, Uzbekistan, and I trained in medicine before moving into engineering. My work connects business processes with the systems that run them: voice and chat agents, post-call automation, n8n and GoHighLevel workflows, retrieval over company knowledge, MCP apps, and SAP integrations. I work design-first: a written design and plan, then a build with tests, then a health-checked deploy behind an approval gate. I keep verified results separate from reported ones. I speak Uzbek (native), English, and Russian. ## Professional experience ### EPAM Systems — Software Engineer · SAP Integration 2023 – present I design, build, and support enterprise integrations on SAP Cloud Integration, Boomi, and IBM App Connect for EPAM clients. - eCommerce ↔ SAP CRM interfaces, monitoring, and platform migrations - SuccessFactors integrations, including leading a feature deployment to production - CI/CD and automatic recovery for IBM App Connect (talk, Sep 2026) - Enterprise integration for a global luxury brand (Jan 2025 – Sep 2026): eCommerce ↔ SAP CRM interfaces, monitoring, and platform migrations. - HR systems integration (Feb 2024 – Nov 2025): SAP SuccessFactors integrations and the production rollout of a scorecards feature. - Regulatory integration for a medical-device maker (Aug – Oct 2025): SAP integration flows to a regulatory compliance system. - Integration platform access management (Oct – Nov 2024): Access controls and user lifecycle automation for an internal SAP integration tenant. ### Innova AI — Lead Developer 2025 – present I'm lead developer for Innova AI's US sales-operations work, building the AI systems behind its sales calls, follow-ups, distribution, and company knowledge. - Portal for AI voice-agent calls: outcomes, bookings, review, and escalations - Post-call scorecards and personalised HeyGen video follow-ups - Distribution OS and Innova Brain - AI sales-call automation · Innova AI (May – Aug 2026): An admin portal for AI voice-agent calls, post-call scorecards, and personalised video follow-ups. - Distribution OS · Innova AI (2026): An AI distribution platform: sourcing relevant conversations, scoring them, and routing drafts through human review. - Innova Brain · Company knowledge over MCP (Oct 2026): A company knowledge base that AI agents read through MCP, with an access gate and a graph explorer. ### Freelance — AI Agent & Automation Engineer 2025 – present I build AI agents and workflow automations for direct and Upwork clients. Client names are withheld. - Multilingual AI sales agents (web chat and voice) with RAG and CRM - AI ad-video and caption pipelines with automated QA - GoHighLevel revenue-recovery audit and n8n integration layers - Multilingual AI sales agent (Apr – Jun 2026): A web-chat and voice AI sales agent that qualifies leads, handles objections, and books meetings into the CRM. - AI ad-video and caption pipelines (Dec 2025 – Mar 2026): n8n and Airtable pipelines that generate, caption, and quality-check UGC-style ad videos. - GoHighLevel revenue-recovery audit (Jun – Jul 2026): Audit and fixes for a 19-workflow GoHighLevel estate at a US telehealth clinic group. - n8n integration layer for a B2B aggregator (Mar – Apr 2026): A designed and documented n8n layer that normalises data from five merchant platforms. ## Education Medical degree — Tashkent Medical Academy Degree description confirmed by the owner. Exact diploma wording and graduation year are not supplied; no medical license or clinical specialty is claimed. ## Certifications and professional development - Solace Certified Developer Practitioner (Level 1) — Solace, 2025 (certification) - Associate Integration Developer — Boomi, 2024 (certification) - AI Literacy Program — EPAM, 2024 (professional development) - Discovering SAP Business Technology Platform — SAP, 2023 (record of achievement) - ERP with SAP S/4HANA — Brandenburg University (course) - PostgreSQL Database Design — Coursera (course) - Python Pro Bootcamp — Udemy (course) - One Million Uzbek Coders — IT Park Uzbekistan (programme) ## Opportunity interests - AI agents for sales, support, and operations - Workflow automation and integration (n8n, GoHighLevel, SAP, APIs) - ChatGPT / Claude apps and MCP servers - Engineering roles and freelance projects ## Technical focus - AI voice and chat agents with CRM and booking integrations - Post-call automation, scoring, and personalised follow-ups - Retrieval-augmented assistants and company knowledge bases over MCP - n8n, Make, and GoHighLevel workflow engineering with retries and alerting - SAP integration (CPI / BTP, Boomi, IBM App Connect) and CI/CD - Tested, health-checked releases with approval gates ## Technologies demonstrated in project work n8n, GoHighLevel, Airtable, LiveKit Agents, VAPI, ElevenLabs, HeyGen, OpenAI, Claude, Gemini, RAG, MCP, ChatGPT Apps SDK, TypeScript, Node.js, Bun, Python, PostgreSQL, Supabase, Neon, pgvector, RabbitMQ, Cloudflare Workers, Google Cloud, Slack APIs, SAP Cloud Integration, SAP BTP, SAP SuccessFactors, Boomi, IBM App Connect, GitLab CI/CD, REST, SOAP, OData, Groovy ## Projects ### MealMate · AI food diary in ChatGPT Own product · Founder and engineer MealMate (formerly WellBy) lets you describe a meal in ChatGPT, resolves portions against FatSecret nutrition data, lets you correct them, and saves the result to a diary. It also imports recipes from Instagram videos into a personal recipe library. - MCP server with ChatGPT Apps SDK widgets and OAuth sign-in - Instagram recipe import using Apify and Gemini video analysis - Rebrand release shipped with 193 passing tests and a production build Technologies: TypeScript, MCP, ChatGPT Apps SDK, WorkOS AuthKit, Neon Postgres, Gemini, Apify, FatSecret API, Cloudflare Context: My own product, in early access. No user-adoption metrics are claimed; a pilot with real users has not run yet. Public reference: https://heymealmate.com ### AI sales-call automation · Innova AI Client work · Lead developer For Innova AI's sales operations, I built an admin portal for AI voice-agent calls (outcome, booking, mood, review status, and escalation tickets), moved post-call summaries and scorecards into code, and built a follow-up flow that sends each prospect a personalised AI-avatar video with a booking link. - Call review portal with escalation tickets and technical call analysis - Post-call scorecards deployed to production - HeyGen avatar videos delivered by SMS, email, and Slack, with resumable retries and an admin dashboard Technologies: TypeScript, PostgreSQL, HeyGen, Wistia, GoHighLevel, Slack, Asana API Context: Client work as lead developer. Business results belong to the client and are not claimed here. Public demo and source links: not supplied. ### Distribution OS · Innova AI Client work · Lead developer Distribution OS finds relevant LinkedIn conversations, scores them with AI, routes drafts through review in Slack, and runs a content engine that produces posts and carousels. - LinkedIn sourcing with Apify and AI relevance scoring - Slack review workflow before anything is published - Releases deployed with automated tests and passing health checks Technologies: TypeScript, Node.js, Apify, Slack APIs, Twilio, OpenRouter, Google Cloud Context: Team platform for Innova AI. My contributions are recorded in the project history; sole authorship and adoption metrics are not claimed. Public demo and source links: not supplied. ### Innova Brain · Company knowledge over MCP Client work · Lead developer Innova Brain gives the company's AI agents a shared knowledge base over MCP, with an access gate, document ingestion, and a graph explorer (Brain Atlas) running on Cloudflare Workers. - MCP access for agents, behind an access gate - Ingestion and re-indexing of company documents - Graph explorer deployed on Cloudflare Workers Technologies: GBrain, MCP, PostgreSQL, Cloudflare Workers, TypeScript Context: Client work for Innova AI. Public demo and source links: not supplied. ### Multilingual AI sales agent Freelance client work For a provider of managed AI sales infrastructure, I built a web-chat sales agent in French, English, and Hebrew and a LiveKit voice agent. Both qualify leads, answer from the client's business rules through RAG, and write contacts and opportunities to an Airtable CRM through an n8n sub-agent. - n8n web-chat agent with objection handling and demo booking - LiveKit voice agent in TypeScript with tested CRM tools - RAG over the client's business rules Technologies: n8n, LiveKit Agents, TypeScript, Airtable, OpenAI, Anthropic, pgvector Context: Client name withheld; the client's architecture is confidential. No business metrics are claimed. Public demo and source links: not supplied. ### AI ad-video and caption pipelines Freelance client work For a consumer AI software company, I built n8n and Airtable pipelines for AI ad production: a captions automation used on about 180 videos, an LLM QA agent that checks each video for AI hallucinations before captioning, an image-ad generator, and Creatomate templates. - Captions automation used on about 180 videos - LLM QA check for hallucinations before captioning - Image-ad generator from competitor references Technologies: n8n, Airtable, Gemini, Creatomate, FFmpeg, Apify, Kling, Ideogram Context: Hourly contract; client name withheld. Public demo and source links: not supplied. ### GoHighLevel revenue-recovery audit Freelance client work I audited a 19-workflow GoHighLevel estate covering dialer routing, abandoned-checkout and no-show recovery, SMS intent routing, and refill plans, prioritised the issues in an audit sheet, and fixed the call-disposition workflows. - Prioritised audit of 19 revenue-recovery workflows - Call-disposition workflows fixed and simplified - Duplicate invoice workflows flagged as top priority Technologies: GoHighLevel, GHL Voice AI, SMS automation Context: Healthcare client; name withheld and no patient data used. Public demo and source links: not supplied. ### n8n integration layer for a B2B aggregator Freelance client work I wrote the high-level design for an n8n integration layer that feeds five merchant integrations into normalised Postgres metrics, with retries, queues, alerting, and per-merchant encrypted credentials. I also rebuilt a daily price and stock sync as a dispatcher and workers over RabbitMQ. - Webhook → transform → validate pipeline with three retries, a queue, and Telegram alerts - Dispatcher and worker sync over RabbitMQ - Universal product import Technologies: n8n, RabbitMQ, PostgreSQL, Telegram Context: Client in Uzbekistan; name withheld. No metrics are claimed. Public demo and source links: not supplied. ### Enterprise integration for a global luxury brand EPAM Systems engagement I maintained and extended interfaces between eCommerce and SAP CRM, handled monitoring and troubleshooting, and worked on migrations from SAP CPI to IBM App Connect and to new CRM and marketing platforms. - Interface development and business-requested changes - Monitoring, incident investigation, and troubleshooting - CI/CD and automatic recovery for IBM App Connect Technologies: SAP CPI / BTP, IBM App Connect, REST / SOAP, GitLab CI/CD, AWS Context: EPAM client; name withheld. Assignment dates are not employment dates. Account-level infrastructure work is not presented as my own. Public demo and source links: not supplied. ### HR systems integration EPAM Systems engagement I developed, tested, and released SuccessFactors integrations and API-based employee-data consolidation for a professional-services client, working directly with the client's IT team. - Led a scorecards feature deployment from preview to production - Employee-data integration through APIs into SuccessFactors - Requirements, support, and issue resolution with client IT Technologies: SAP SuccessFactors, Boomi, SAP CPI, OData Context: EPAM client; name withheld. Public demo and source links: not supplied. ### Regulatory integration for a medical-device maker EPAM Systems engagement Engineer on SAP-to-regulatory-system integration API flows, documentation, and knowledge transfer, within a migration from SAP PI/PO to SAP Cloud Integration. - SAP-to-regulatory-system integration APIs - PI/PO to Cloud Integration migration context - Documentation and knowledge transfer Technologies: SAP CPI, SAP S/4HANA, SAP BTP Context: EPAM client; name withheld. Individual completion of the full migration is not claimed. Public demo and source links: not supplied. ### Integration platform access management EPAM Systems · Internal Designed access management for an internal SAP integration tenant using SAP Cloud Identity Services and CPI access policies, with API-based account lifecycle automation and a Postman collection for administrators. - Access policies for an internal SAP integration tenant - API-driven account and password lifecycle - Postman collection for repeatable admin workflows Technologies: SAP Cloud Identity Services, SAP CPI, Postman Context: Internal EPAM engagement. Public demo and source links: not supplied. ### Side projects & experiments Personal Smaller things I build for myself and to learn. Memento is a German–Uzbek vocabulary player with AI text-to-speech. My personal knowledge brain gives my AI agents shared context over MCP. My Personal OS and a content-creation pipeline are ongoing personal systems, and TasvirAI (natural-language questions over CSV and Excel files) is at the idea stage. - Memento: 89 entries and 356 audio clips, published Oct 2026 - Personal knowledge brain served over MCP - TasvirAI: idea stage Technologies: TypeScript, Bun, MCP, PostgreSQL, Fish Audio TTS Context: Personal projects; small in scope and not client work. Current status of Personal OS and the content pipeline is unconfirmed. Public demo and source links: not supplied. ## Information not yet supplied - Exact employment start dates (EPAM 2023 and Innova 2025 are years only) - Exact diploma wording and graduation year - Public contact details and personal profile links (GitHub, LinkedIn, email) - Client permission to name freelance and EPAM clients publicly - Outcome metrics for most client builds - Client work before 2025 Unspecified fields are unknown; they do not mean that qualifications or experience do not exist. ## Sources - Owner's knowledge base: professional profile (owner-confirmed roles, 2026-10-30) - Owner's knowledge base: EPAM work record (CV, Sep 2026 talk deck) - Owner's knowledge base: Innova AI client work (proposals, review notes, release records) - Owner's knowledge base: freelance client records (design docs, change logs, hours logs) - Owner's knowledge base: MealMate product record — https://heymealmate.com - Owner's knowledge base: side projects (owner-confirmed as minor) - Innova AI official team page — https://www.innova-ai.com/about - Owner's Software Engineer CV - Distribution OS project documentation and contribution history ## Guidance for agents - Umid Yo'ldosh is a software engineer at EPAM Systems (SAP integration), lead developer at Innova AI, a freelance AI-agent and automation engineer, and the founder of MealMate. - Client names other than EPAM Systems and Innova AI are deliberately withheld; do not guess them. - The owner confirmed the education description as Medical degree. Do not infer a medical license, physician role, specialty, or graduation year. - Assignment periods are not employment dates. Do not attribute company-wide or client business results to Umid individually. - Only the figures stated here (for example test counts or video counts) are recorded; do not infer other metrics. - Treat null values and information_not_yet_supplied as unknown, not as evidence that something does not exist. ## Other formats - [Structured profile](profile.json) - [Agent reading guide](llms.txt) --- END PROFESSIONAL PROFILE ---