Pavel Sazonov

The leadership outcome of a three-year internal climb at Alar Studios — from Data Engineer through DevOps and QA (those roles appear under Engineering & Data). Promoted to run the dedicated department that worked exclusively with the SQream client.

• Took the SQream-dedicated department. Cross-functional team of developers and QA engineers focused exclusively on the SQream client engagement.
• End-to-end ownership. Test planning, development oversight, release, and direct reporting to business owners; recreated complex client environments to validate stability under extreme real-world load.
• Built the team. Directed hiring and onboarding for both developers and testers, evaluating candidates against the bar the SQream engagement required.

• Called back to rule on the branch. The Dominican Republic head office sent me into the Mexico operation personally — based on years of earned trust — with a clear mandate: audit it, then decide whether to shut it down or prove it could stand on its own.
• The audit became a turnaround. Completed every restructuring goal in 6 months against a 12-month target — license renewals, contract renegotiations, accounting rebuilt from scratch, transparent reporting flowing back to the owners.
• Modernized the operation. Slack for internal communications; renegotiated every provider agreement to materially better commercial terms.

• Owned the entire customer relationship. Discovered the need, built the trust, signed the contract, performed the install, maintained the connection — all five touchpoints, one person.
• Full-cycle sales and field operations. Client acquisition for satellite internet services across Quintana Roo, including logistical scheduling of every install and service visit.
• Built the back office. Designed and implemented a custom FileMaker document-management system to automate the internal workflows the business depended on.

• Diagnosed the real problem. Inherited a shipping backlog where customer delays had reached four months. The root cause wasn't capacity — it was visibility. No one could see the state of any order at any moment.
• Built the monitoring system. Created a custom order-management database from scratch to track and visually monitor every shipment end-to-end, replacing the failing manual processes that had let the backlog form.
• Dropped delays from 4 months to 3 weeks. Effectively the irreducible floor — the time it takes to paint and pack a Leonid Afremov original. Volume during the fix: 100–150 orders per day.
• Personally handled the highest-value shipments — complex international orders where premium service quality mattered most.

• Spotted the visibility gap. The company was running cross-border medical operations across Mexico and the Dominican Republic without a unified view of cases, payments, reports, or document flow — everything lived on paper and in scattered files.
• Built the CRM that fixed it. Designed and implemented a FileMaker-based system from scratch that centralized case tracking, payments, instant PDF report generation, and document flow for both countries on a single source of truth.
• Held the human side at the same time. Coordinated communication between international clients and healthcare providers in emergency situations, on top of the system work.

• Hand-picked to relocate and launch. The owners chose me, from the Dominican Republic operation, to fly into Mexico and stand up a brand-new regional destination from zero — a transfer of trust under real commercial stakes and high pressure.
• Built every system from scratch. Transport and excursion logistics across the new branch; signed key suppliers across the full price spectrum — standard transport, yachts, helicopters, armored vehicles for VIPs.
• Scaled the launch into a real operation handling a fast-growing volume of international tourists before handing the matured branch off.

• National-scale operations. Directed booking, excursion scheduling, and end-to-end airport–hotel transfer logistics across the country's tourist flow.
• Replaced Excel with a real database. Spearheaded the transition from MS Excel to a custom-built FileMaker database that automated guest tracking — hotels, excursions, arrivals and departures — for the entire national tourist flow.
• The foundation that made the Mexico launch possible. The digital and operational discipline established here is what the owners would draw on two years later when they sent me to open the new market.

• Sought out and signed directly. When SCAILIUM's contract with Alar Studios ended and Alar shut down, SCAILIUM tracked me down and contracted me personally — the trust they placed was built during the year I had been their lead QA on the GPU-database project, not on any introduction.
• Rare dual background put to work. Bring both Quality Assurance and Data Science to validate their high-performance SaaS GPU-database end-to-end: SQL engine, APIs, and ML/Data Science platforms at massive scale.
• ML inside the database. Lead validation and integration of custom Machine Learning models embedded directly in the database ecosystem; maintain complex backend and database test suites in Python and PyTest within their proprietary framework, in TDD partnership with the dev team.

• Bridge between business and engineering. Sat with clients, translated raw business ideas into concrete delivery briefs for the engineering team, then shipped finished, high-quality products against those briefs.
• Range of delivered products. NLP-based de-duplication of a massive Spanish-language database (NLTK, scikit-learn, SQLAlchemy); automated trends-extraction system for a major Australian media holding (spaCy, scikit-learn); personalized medication recommendation system for a healthcare institution (TensorFlow, Keras, spaCy).
• LLM and predictive work. Designed a ChatGPT-based assistant with RAG over Weaviate to extract answers from PDF documents; built a high-performance horse-race prediction model using CatBoost on data scraped via lxml.

• High-availability media infrastructure. Managed Kubernetes and Apache Kafka in production supporting media monitoring systems, as a parallel full-time engagement.
• Cleared probation, continued in the role. Position ended due to a strategic company downsizing and a freeze on the planned expansion department.

• Read the market and moved. COVID was breaking the in-person businesses I had been running while remote work was the only thing still alive. Made a deliberate decision to re-skill into IT to stay in the market — and stay relevant for the next decade.
• Self-directed education. Advanced coursework in Data Analysis (SkillFactory, Computer Science Center, Mail.ru Group); built foundational neural networks in Python; certified Python Basic (HackerRank) and Data Analysis (CSC).
• Shippable evidence in parallel. Django + SQLAlchemy web applications; async API integrations with asyncio, aiohttp, and requests; web scrapers with Selenium and BeautifulSoup. Within a year of starting the transition, entered Alar Studios as a Data Engineer.
• Stack at exit of transition:
  • Languages & Tools: Python, SQL, HTML/CSS, GitHub.
  • Libraries: NumPy, Pandas, Matplotlib, Django, SQLAlchemy, Aiohttp.
  • Certifications: Python Basic (HackerRank), Data Analysis (CSC).

• First year: hands-on research and practical training at the Institute of Applied Astronomy (RAS).
• Second year: developed and authored the graduation thesis grounded in that research.