NayaOne · 2023–2024
Killing a GraphQL N+1 that threatened Tier 1 bank renewals
Rewrote 80+ GraphQL resolvers with raw SQL and cross-database batching; hot paths went from 250ms to 75ms.
- 80+ queries refactored
- 250ms → 75ms (−70%)
- 300+ daily users
Egor Kravchenko
Senior Full-Stack & Mobile Engineer
I build and stabilize TypeScript systems for fintech and healthcare — from backend APIs to React Native apps.
Antalya, Turkey · Remote only · UTC+3
Open to senior / staff roles · full-time or contract
Selected outcomes
Selected work
All projects →Medical Avenue · 2026
Taking over a React Native app that was OOM-killing itself: 1.2GB → 350MB
Traced an inherited app's 1.2GB memory climb to its navigation architecture and rebuilt it to a stable 350MB.
- 1.2GB → ~350MB stable
- OOM kills on devices → zero
- Verified on 2018 Android, 3GB RAM
Medical Avenue · Dec 2025 – Jan 2026
A RAG system built in 2.5 weeks to walk into a CEO meeting with proof
Source-grounded RAG over 264 videos, 2 books and 7 papers, built solo in 2.5 weeks before any contract existed.
- 264 videos + 2 books + 7 papers ingested
- Cited, source-grounded answers only
- ~2.5 weeks, solo, pre-contract
NayaOne · 2025
FCA–NVIDIA AI Sandbox: a launch feature shipped in one week
Shipped the FCA–NVIDIA AI Sandbox launch module in one week, dark behind feature flags on a live platform.
- 1-week delivery deadline
- Launched on schedule — June 2025
- Official FCA technology partner
Working on a system like these? Message me on LinkedIn →
Featured writing
View all →- 20 Aug 2026How I stopped re-explaining the project to every AI agentDecisions lived in chats, then in two docs folders, then in one knowledge base that Claude Code searches by meaning. The episode that convinced me, and the rule that came with it: a found note is a claim, not a fact.
- 26 Jul 2026Four AI reviewers, one schema, and a fix I revertedI had four models audit the same Prisma schema twice each, verified every claim by script, and still implemented a critical finding the application had already solved. What eight AI reviews are good for, and what they are not.
- 05 Jul 2026Mapping a legacy Prisma schema without reliable telemetryMore than 150 models, legacy pages still in production, and traffic data I could not trust. How I worked out which models a limited set of functions actually depends on, where the script needed a human, and what the result does not prove.
- 15 Jun 2026I built an AI agent that knows meA self-hosted AI agent with local embeddings, hybrid retrieval, and persistent memory over my own knowledge base — and why the hard part wasn't the model.