Founder. Researcher. Builder.

Vinayak Sharma

AI/ML Engineer • Research Builder • Polymath

I build full-stack web applications, decentralized products, and intelligent systems with a strong focus on Web Development, Cryptography, Blockchain, and AI/ML. Since 2020, I’ve delivered 28+ projects across diverse domains and continue to explore the frontier of technology through product building, hackathons, and research-driven experimentation.

10+

Research-first builds

AI / Systems / Web3

Core domains

Ship fast, explain deeply

Focus

Research-paper implementationsProduction-minded backend systemsCrypto-native MVPsTechnical storytelling for demos

Selected Work

A curated mix of research-faithful AI implementations, enterprise workflow systems, and crypto-native product experiments.

CLI-Network Tool2026

Truss

truss is a Linux failover network manager CLI written in Rust. It continuously measures internet health on the active uplink using interface-bound ping probes, and it switches the default route to a backup uplink when the primary becomes degraded or dead.

Rustrata-uiUNIX
Live Prototype
Explore project
Deep Learning Research2025

Music-Recommendation System

A paper-based implementation and comparative study of CNN-ORNN-SIAO, CNN-LSTM-SIAO, and SSA-CNN-LSTM for safety-critical fault diagnosis using simulated plant data and reliability-oriented analysis.

PythonTensorFlowKerasNumPy
Research Implementation
Explore project
Crypto / Bots2025

Telegram On-chain Analytics Bot

A crypto community bot using Vybe APIs to deliver wallet tracking, whale alerts, token metrics, and on-chain signal summaries directly in Telegram for community-native alpha workflows.

PythonTelegram Bot APIVybe APIsWebhooks
In Progress
Explore project

Research Mindset

I prefer technically honest work: reproduce the baseline, verify the system, isolate the innovation, and then present the delta clearly.

Baseline Fidelity

I start by implementing the original paper architecture as closely as possible. This creates a trustworthy baseline and prevents fake novelty.

Controlled Improvement

Once the baseline is stable, I change one meaningful component at a time so the comparison remains interpretable and defensible.

Evaluation Beyond Accuracy

I care about precision, recall, reliability, execution trade-offs, and how the model behaves in a practical deployment narrative.

Presentation-Ready Output

Every project is shaped into a demo, a story, and a deployable artifact—not just a notebook result.

Technical Stack

A practical toolkit across machine learning, backend engineering, frontend systems, and web3 experimentation.

AI / ML

PythonTensorFlowKerasPyTorchScikit-learn

Backend

DjangoFastAPIREST APIsPostgreSQLRedis

Frontend

ReactNext.jsTailwindTypeScript

Web3 / Infra

SolanaStarknetMonadDockerAWSGitHub Actions

Latest Writing

Notes on AI systems, Linux internals, research-to-product workflows, and crypto build strategy.

Research5 min read

How I Turn Research Papers into Working AI Projects

My practical framework for converting academic papers into demos, comparisons, and presentation-ready engineering artifacts.

2026-04-10

Read post