101 lines
5.5 KiB
TypeScript
101 lines
5.5 KiB
TypeScript
export const resume = {
|
||
name: 'Emil Shanayev',
|
||
eyebrow: 'AI systems · scientific software · infrastructure',
|
||
role: 'AI Systems & Scientific Software Developer',
|
||
bio: 'I build technical products at the intersection of scientific computing, AI-native software and systems engineering — from molecular similarity tools to a Vulkan renderer and production-oriented backend infrastructure.',
|
||
location: 'Russia',
|
||
availability: 'Open to internships, research collaborations and engineering projects',
|
||
contacts: [
|
||
{ label: 'GitHub', value: 'emil28092005', href: 'https://github.com/emil28092005' },
|
||
{ label: 'Email', value: 'emil28092005@gmail.com', href: 'mailto:emil28092005@gmail.com' },
|
||
{ label: 'Email', value: 'emil28092005@yandex.ru', href: 'mailto:emil28092005@yandex.ru' },
|
||
{ label: 'Telegram', value: '@emilshanaty', href: 'https://t.me/emilshanaty' },
|
||
],
|
||
capabilities: [
|
||
{
|
||
index: '01',
|
||
title: 'Scientific software',
|
||
text: 'Molecular data processing, cheminformatics workflows, numerical methods and reproducible command-line tools.',
|
||
stack: ['Python', 'RDKit', 'ChEMBL', 'NumPy', 'Flask'],
|
||
},
|
||
{
|
||
index: '02',
|
||
title: 'AI-native systems',
|
||
text: 'MCP integrations, local model tooling, persistent model state and human-in-the-loop workflows.',
|
||
stack: ['MCP', 'LLM APIs', 'PyTorch', 'SSM / Mamba', 'Computer Vision'],
|
||
},
|
||
{
|
||
index: '03',
|
||
title: 'Systems & infrastructure',
|
||
text: 'Performance-oriented engines, backend services and the infrastructure needed to ship and operate them.',
|
||
stack: ['C# / .NET', 'Vulkan', 'Go', 'Docker', 'Jenkins', 'Linux'],
|
||
},
|
||
],
|
||
projects: [
|
||
{
|
||
number: '01',
|
||
name: 'Cortex Engine',
|
||
kind: 'AI-native 3D engine',
|
||
description: 'A 3D engine with a pure P/Invoke Vulkan 1.3 renderer, PBR shading, cubemap-array shadows, Jolt physics and Flecs ECS. An MCP interface exposes scene operations to AI agents, including object creation, transforms and screenshots.',
|
||
highlights: ['Pure Vulkan 1.3 backend', 'PBR + multi-light shadows', 'Playable FPS demo', '7 MCP scene tools'],
|
||
tech: ['C#', '.NET 9', 'Vulkan', 'SDL3', 'Jolt', 'Flecs', 'MCP'],
|
||
href: 'https://github.com/emil28092005/Cortex_Engine',
|
||
},
|
||
{
|
||
number: '02',
|
||
name: 'SciMesh',
|
||
kind: 'Scientific computing framework',
|
||
description: 'A local Python framework for exact molecular similarity search and sparse similarity-graph construction over ChEMBL-style datasets. It streams input data, bounds top-k memory use and avoids dense N×N similarity matrices.',
|
||
highlights: ['Morgan fingerprints', 'Exact Tanimoto search', 'Block-wise graph construction', 'Tested CLI package'],
|
||
tech: ['Python', 'RDKit', 'ChEMBL', 'Pytest'],
|
||
href: 'https://github.com/emil28092005/SciMesh',
|
||
},
|
||
{
|
||
number: '03',
|
||
name: 'Bioportal Lab',
|
||
kind: 'Laboratory CV prototype',
|
||
description: 'A Russian-language prototype for microbiology workflows. PetriCount combines multi-scale computer-vision candidates, a Random Forest classifier and manual correction to support colony counting from Petri-dish images.',
|
||
highlights: ['Image upload workflow', 'Trainable CV baseline', 'Manual result correction', 'Local analysis API'],
|
||
tech: ['Python', 'Flask', 'OpenCV', 'scikit-learn', 'JavaScript'],
|
||
href: 'https://github.com/emil28092005/Bioportal-lab',
|
||
},
|
||
{
|
||
number: '04',
|
||
name: 'Shagram',
|
||
kind: 'Backend & DevOps system',
|
||
description: 'A real-time WebSocket chat used as an end-to-end infrastructure project: JWT authentication, SQLite persistence, containerized delivery, TLS termination and a Jenkins pipeline connected to a private Harbor registry.',
|
||
highlights: ['WebSocket rooms', 'JWT + origin validation', 'Multi-stage containers', 'Automated CI/CD on VPS'],
|
||
tech: ['Go', 'Gin', 'WebSocket', 'Docker', 'Nginx', 'Jenkins', 'Harbor'],
|
||
href: 'https://github.com/emil28092005/shagram',
|
||
},
|
||
{
|
||
number: '05',
|
||
name: 'Memba + memwalk',
|
||
kind: 'Experimental model memory',
|
||
description: 'An experimental stack for saving and restoring persistent state in SSM-based language models. memwalk builds on memba to digest codebases once, cache model state and answer later questions without repeatedly ingesting the same repository.',
|
||
highlights: ['Portable state snapshots', 'C and Python APIs', 'Manifest-based cache invalidation', 'MCP server for codebase queries'],
|
||
tech: ['C++', 'Python', 'CMake', 'llama.cpp', 'Mamba / SSM', 'MCP'],
|
||
href: 'https://github.com/emil28092005/Memba',
|
||
secondaryHref: 'https://github.com/emil28092005/memwalk',
|
||
},
|
||
],
|
||
tools: [
|
||
{ name: 'Soundgen', note: 'Rust-based 8-bit sound synthesis and feedback tooling', href: 'https://github.com/emil28092005/Soundgen' },
|
||
{ name: 'KimiFleet', note: 'Visible multi-agent orchestrator over ACP and MCP', href: 'https://github.com/emil28092005/KimiFleet' },
|
||
{ name: 'Imagen', note: 'Local text-to-sprite MCP pipeline with rating workflow', href: 'https://github.com/emil28092005/Imagen' },
|
||
{ name: 'MoME', note: 'Sparse-gated personal memory plugin for Hermes Agent', href: 'https://github.com/emil28092005/hermes-plugin-mome' },
|
||
],
|
||
education: [
|
||
{
|
||
period: '2023—2025',
|
||
institution: 'Innopolis University',
|
||
program: 'Data Structures and Artificial Intelligence (DSAI)',
|
||
},
|
||
{
|
||
period: '2026—present',
|
||
institution: 'Central University',
|
||
program: 'Currently studying',
|
||
},
|
||
],
|
||
} as const
|