Senior Full-Stack Engineer / Lev

6+ years of experienceBelgrade, Serbia

Senior full-stack engineer with 6 years of commercial experience, the last 4+ years on a single high-traffic mortgage platform - promoted from junior to senior and now the platform's largest contributor to its GraphQL BFF. Led a full legacy rewrite that cut feature delivery time 3–4×, and personally built the platform's production AI features, including an in-house document recognition system that replaced a €144k/year vendor. Deep TypeScript/React core with production Go and Python experience, plus a strong independent track record shipping AI products end-to-end.

Skills

  • Fast, proven growth: Promoted junior → senior within the same company during a major platform overhaul - direct signal of trust and impact, not just tenure.
  • Core GraphQL/TypeScript ownership: Largest contributor to the platform's GraphQL BFF (Node.js/TypeScript over gRPC microservices) - schema design, service aggregation, auth propagation at scale.
  • High-impact rewrite: Led the rewrite of a legacy 7-page, hundreds-of-fields mortgage application into idiomatic React (React Query, React Hook Form), cutting feature delivery time 3–4× and reducing bugs.
  • Real production AI/LLM experience: Built a self-hosted document recognition system (Gemma via llama.cpp) that cut bank document rejection from ~45% to 1–5% and replaced a €144k/year vendor - plus co-built a multi-agent RAG assistant used by 100+ employees daily.
  • Broad technical range: TypeScript, React, Next.js, Go, Python (FastAPI), GraphQL, gRPC, PostgreSQL/pgvector, Redis, MongoDB - strong frontend-to-backend versatility.
  • Independent builder mindset: Ships production-grade side projects solo (a live AI travel planner with ~9,600 monthly visitors, a published RAG evaluation system) - shows initiative well beyond day-job scope.

Experience

Jan 2022 — Present

Senior Full-Stack Engineer (AI Features)

M2 (Mortgage Services)
  • Largest contributor to GraphQL BFF (Node.js/TypeScript) over gRPC microservices — schema design, service aggregation, auth propagation
  • Led rewrite of legacy mortgage application (7 pages, hundreds of fields) to idiomatic React with React Query and React Hook Form — cut feature delivery time 3–4× and reduced bug rate
  • Designed and implemented integrations with external bank APIs and government services for automated data retrieval and submission
  • Shipped in-house AI document recognition system (self-hosted Gemma 26B via llama.cpp) — cut bank rejection from ~45% to 1–5%, replaced €144k/year vendor, 300–1,000 docs/day at 95–99% accuracy
  • Built multi-agent RAG assistant serving 100+ employees at thousands of queries/day — TypeScript orchestration + Python analytics, role-aware answering
  • Team reference for LLM work — mentor senior engineers on RAG, agent design, and evaluation practice
Aug 2020 — Dec 2021

Software Engineer (Full-Stack, Multi-Project)

Petrosoft (Outsourcing)
  • Generalist across 3–4 parallel client projects — TypeScript/React frontends, Java and Python backends (logistics, booking, anti-fraud)
  • Assessed failing project as unworkable, made case for full rewrite, delivered to client production in under two weeks

Education

2015

Specialist Degree, Civil Engineering

Saint Petersburg University of Architecture and Civil Engineering

Contact

Takovska 5, Belgrade, 🇷🇸 Serbia.

Email: [email protected]