Viacheslav Dreval - AI Developer | ContraWork by Viacheslav Dreval
Viacheslav Dreval

Viacheslav Dreval

Java Architect | Cloud, DevOps & AI/LLM

New to Contra

Viacheslav is building their profile!

Cover image for Self-Hosted AI/LLM & RAG Platform

Designed
Self-Hosted AI/LLM & RAG Platform Designed and built a self-hosted AI/LLM platform for inference, embeddings, and RAG workloads, with a focus on practical deployment, performance, and integration with production backend systems. Deployed LLM inference using vLLM on NVIDIA GPUs and Ollama, configured GPU-enabled Docker environments, and evaluated CPU/GPU inference performance. Built RAG pipelines using embedding models, PostgreSQL vector search, document processing, and retrieval to provide relevant context to LLMs. Integrated both self-hosted models and managed AI services through AWS Bedrock, exposing AI capabilities to existing backend applications and services. Key technologies: vLLM, NVIDIA GPUs, Docker, Ollama, AWS Bedrock, PostgreSQL, vector search, embeddings, RAG, Python, LLM inference.
0
14
Cover image for GPXLook – Online GPX Viewer
GPXLook – Online GPX Viewer & Tools Built and launched GPXLook, a privacy-first online GPX viewer and toolkit that processes files entirely in the browser without uploading user data to a server. Developed interactive map visualization, elevation and speed analysis, multi-track support, waypoint handling, and tools for merging, splitting, reversing, simplifying, and converting GPX files. Designed the application as a fast, frontend-only product with static SEO-friendly pages and production deployment through Cloudflare. Key technologies: React, TypeScript, Vite, MapLibre, OpenFreeMap, Cloudflare.
0
10
Cover image for ScribeShark — AI Transcription Platform

Built
ScribeShark — AI Transcription Platform Built and deployed an AI-powered transcription platform for processing audio and video content. Implemented the backend and production infrastructure, including speech-to-text processing with faster-whisper, API services, containerized deployment, and reverse proxy configuration. Designed the system to run AI workloads efficiently on self-managed infrastructure while keeping the architecture ready for GPU acceleration and scaling. Key technologies: Python, FastAPI, faster-whisper, Docker, Nginx, Linux, AI/ML, speech recognition.
0
5
Cover image for 12ozCode — Independent Software Products

Built
12ozCode — Independent Software Products Built and launched an independent software portfolio focused on practical developer and desktop tools. Designed product architecture, websites, deployment pipelines, release workflows, and distribution infrastructure for multiple applications. Projects include native Windows desktop utilities built with Rust/Tauri and C#/.NET, alongside web applications deployed through Cloudflare. Implemented automated builds, installers, release packaging, monitoring, analytics, and secure software distribution. Key technologies: Rust, Tauri, C#/.NET, TypeScript, Cloudflare, CI/CD, GitHub, Windows.
0
25