Sargis
Nersesyan

• Built a predictable delivery process for the backend stream (19 people): sprint say/do rose from 55% to 87% and lead time fell by 57% through flow metrics and WIP limits.
• Led the backend side of the US launch under the Fledge brand: a separate production environment and a split between CIS and global users; the app is live on Google Play.
• Moved the backend to daily releases, cutting the change failure rate by 73% and time to recover (MTTR) by 89%.
• Led the extraction of 12 services from the monolith into microservices (Python, Go) with zero downtime.
• Brought AI tools into the development process (code review, autotests), cutting MR review time by 70% and catching 35% of defects before manual QA.
• Led the backend side of key launches: monetization (purchases, promo codes, Faster Payments), the ads manager and AI features (recommendations, filters, message translation).

Stack: Python/Django, Celery, Go, CockroachDB, NATS, ClickHouse, GitLab CI, Yandex Cloud.

• Ran the program retiring 18 outdated technologies across 132 systems: closed the annual plan in 6 months and cut infrastructure rental costs by 49.7% (~RUB 143M).
• Led the launch of an internal AI-agent development platform (Sber's AI-Disrupt PDLC methodology), automating part of the teams' routine work.
• Introduced Specification-Driven Development in pilot teams, cutting time to market from 105 to 45 days.
• Turned 60+ policies into a single “NFR as code” registry: 170 requirements covering about 119 of 132 systems. Automated compliance checks: 90% of deviations are found with no people involved, and fix tasks reach teams automatically.
• Cut incidents in Megamarket's critical services by 54% in six months by stopping the misuse of outdated technologies.

• Managed a portfolio of 30+ change requests worth RUB 95M+: 12.5% budget savings and under 10% schedule variance.
• Introduced OKRs, KPIs and performance reviews for a 25-person team: engagement up 25%, attrition down 20%.
• Streamlined planning and development: team velocity up 20%, key tasks reaching release 15% faster.
• Improved code quality and product stability by introducing code review, unit and integration testing and regular architecture audits.
• Prevented potential losses of RUB 17.8M across 10+ projects through proactive risk management.

• Delivered 2 NLP/ML projects for enterprise clients (MTS, a bank): up to 80% of requests handled automatically, CSAT at MTS up 25%.
• Closed 13 of 14 projects on time and within budget in an outsourcing model.
• Ran 50+ customer development interviews to gather and formalize requirements, cutting rework by 35%.
• Led CRM integrations with 8+ external services (API, payment systems), cutting the time to connect a new service from 2 weeks to 3 days.
• Managed an SAP PP/MM rollout with industrial IoT integration, cutting equipment downtime by 12%.