About

Builder Profile

I build AI infrastructure, language technology for underserved markets, and security systems at production scale. 20+ shipped products, 85+ public repositories, and multiple active ventures in deep-tech AI. My work spans LLM inference engines, on-device ML models, code security agents, privacy-first platforms, and open-source tools used by engineers worldwide.

Core Expertise

Languages & Frameworks

PythonTypeScriptJavaScriptReactNext.jsNode.jsPostgreSQLMongoDBJavaC++RustSQL

AI & Data

PyTorchTensorFlowscikit-learnOllamaHugging FaceLangChainOpenCVNumPyPandasMistral AIGemini API

Infrastructure

DockerLinuxAWSGitHub ActionsNginxBashKubernetesVercelNetlify

Specializations

AI InfrastructureLanguage AIOn-Device MLSecurity EngineeringDeveloper ToolingOpen SourceProduct EngineeringPrivacy Engineering

Focus Areas

AI & Language Infrastructure

Building LLM inference engines, on-device ML models, tokenizer optimization, and language AI for underserved markets.

Security & Privacy

Developing code security agents, on-device threat detection, forensic tooling, and privacy-first systems.

Product Engineering

Shipping production systems from concept to deployment — full-stack, AI-powered, and built to scale.

Open Source & Community

85+ public repositories. Maintaining tools across AI, infrastructure, security, and developer ecosystems.

Who is Sai Dutta Abhishek Dash and what does he build?

Sai Dutta Abhishek Dash is a founder and engineer based in Odisha, India, building AI infrastructure, language technology for underserved markets, and security systems at production scale. He has shipped 20+ production products, maintains 85+ public repositories, and has published 46+ models and 22+ datasets on Hugging Face. His work spans LLM inference engines, on-device machine learning models, code security agents, privacy-first platforms, developer tooling, and 5 AI agent skills. He leads 3 deep-tech AI ventures, including Maelis Research, founded in 2026 to build the first commercial Odia-specific AI API for a population of 38 million speakers. He is a 2025 Computer Science graduate of GIET University Gunupur and an AWS Certified Cloud Practitioner. His advisory practice spans five service lines covering consulting, engineering, and technical due diligence, and he works with partners worldwide from Odisha, India. He ships open source by default.

What are Sai Dutta Abhishek Dash's core technical skills?

Sai Dutta Abhishek Dash's technical skills fall into four groups. Languages and frameworks: Python, TypeScript, JavaScript, Rust, C++, Java, Go, React, Next.js, Node.js, PostgreSQL, and MongoDB. AI and data: PyTorch, TensorFlow, scikit-learn, NumPy, Pandas, OpenCV, Ollama, Hugging Face, and LangChain. Infrastructure: Docker, Linux, AWS, Kubernetes, GitHub Actions, Nginx, Bash, Vercel, and Netlify. Specializations: AI infrastructure, language AI, on-device machine learning, security engineering, developer tooling, open source, product engineering, and privacy engineering. He applies these across 22 shipped production systems, from the C99 ornith-flight inference engine to TypeScript platforms such as Vulscany and Binify, and is AWS Certified as a Cloud Practitioner. He also works with XGBoost, Keras, and Google Vertex for applied modelling and evaluation, and with Turso, NeonDB, and Upstash Redis for data infrastructure. His toolkit spans more than 40 named technologies in total, and he has published 46+ models and 22+ datasets on Hugging Face.

What are Sai Dutta Abhishek Dash's focus areas?

Sai Dutta Abhishek Dash focuses on four areas. AI and language infrastructure: LLM inference engines, on-device machine learning models, tokenizer optimization, and language AI for underserved markets, including the 3x more efficient Odia tokenizer built at Maelis Research. Security and privacy: code security agents such as Vulscany, on-device threat detection such as PhishScout, forensic tooling such as Forensic-Recovery, and privacy-first systems such as Binify. Product engineering: shipping production systems from concept to deployment, full-stack and AI-powered, built to scale. Open source and community: maintaining 85+ public repositories across AI, infrastructure, security, and developer ecosystems, plus 5 AI agent skills. The on-device focus is concrete: PhishScout runs as a 131 KB model with 90% precision at roughly 4 ms per lookup. The security focus ships in five distinct systems, and the open source focus is measured in 85+ public repositories. Product engineering spans Binify, Ansora, and Visitor Analytics.