Neul Labs
Governed agents, durable orchestration, evaluation, local inference, and open systems.
About
Builder first. Researcher because the hard questions deserve evidence.
Across 18+ years, I have taken software, machine-learning products, robotics, and distributed systems from research and architecture into production. My current work focuses on governed agents, durable execution, exact evaluation, local inference, and cloud-to-edge systems.
- years building production systems
- 18+
- user scale for ML systems at Hike
- 100M+
- seeded kernel defects caught in a measured corpus
- 10/10
- companies founded across 2008–2024
- 6
The through-line
From first product to operational AI
I started building software products in the 2000s and founded six companies across social software, interactive media, machine-learning analytics, logistics, blockchain infrastructure, and applied AI. Neul Labs is the current vehicle for my technical and consulting work.
The work then moved through larger operating systems. At Hike I worked on multilingual NLP, recommendation, personalisation, computer vision, and trust-and-safety systems at more than 100M-user scale. At Nykaa I built commerce and data platforms and led engineering teams. At Orangewood Labs I built RobotGPT and worked across computer vision, robotics, and cloud-to-edge systems.
More recently, I architected regulated banking AI at Aveni and delivered agentic systems using Azure AI Foundry and Kubernetes. The recurring question is the same across each environment: what has to be explicit—in the architecture, evidence, authority, and operating model—for people to trust the system with real work?
Research is one way I answer that question. Public papers and systems let other people inspect the methods, reproduce a result, challenge the boundary, and use the work without relying on a biography.
Experience
Systems in different operating environments
Aveni
Enterprise banking AI with evidence, human review, escalation, evaluation, versioning, and release controls; part of the Aveni team in the first FCA Supercharged Sandbox cohort.
Orangewood Labs
RobotGPT, computer vision, collaborative robotics, and cloud-to-edge systems.
Hike
NLP, recommendation, personalisation, vision, and trust-and-safety systems at more than 100M-user scale.
Nykaa
High-throughput commerce, data systems, platforms, and leadership of 15+ engineers.
Manufactured
Python and AWS document extraction, validation, routing, and LLM integration with operational workflows.
Education
M.S. Computer Science (Cybersecurity)
Arizona State University · 2020–2022 · GPA 3.96/4.0
B.Tech. Computer Science & Engineering
Indian Institute of Technology Delhi · 2003–2007
Graduate Certificate in Strategic Studies
The Takshashila Institution · 2018
Books & research
AI for Everyday Automation
Packt · 2026
Nginx 1 Web Server Implementation Cookbook
Packt · 2011
The GenAI and Agentic AI Playbooks
21 chapters · ten languages · Apache 2.0
Applied machine-learning research
Federated learning, AI-generated code, multi-agent engineering, robotics, and decentralised systems.
Direct answers
About Dipankar Sarkar
Who is Dipankar Sarkar?+
Dipankar Sarkar is a founder, fractional AI CTO, and hands-on applied AI systems engineer based in St Andrews, Scotland. Across 18+ years he has built production software, distributed systems, machine-learning products, robotics, and AI platforms.
What is Dipankar Sarkar working on now?+
Through Neul Labs, Dipankar works on governed AI agents, durable execution, evaluation, local inference, AI-generated code correctness, custom sandboxing, and cloud platforms including Azure AI Foundry and Kubernetes.
What is Dipankar Sarkar’s technical background?+
His work spans Python, Rust, Java, Node.js and TypeScript, C and C++, .NET, distributed systems, cloud platforms, model APIs, agent orchestration, data infrastructure, evaluation, and operations.
What research has Dipankar Sarkar published?+
His public work includes 2026 arXiv preprints on GPU-kernel correctness, tensor-program test generation, and multi-agent software engineering, plus Fed-Focal Loss, accepted at FL-IJCAI 2020.
Where did Dipankar Sarkar study?+
He holds an M.S. in Computer Science (Cybersecurity) from Arizona State University and a B.Tech. in Computer Science and Engineering from IIT Delhi. He also completed a Graduate Certificate in Strategic Studies at The Takshashila Institution.
Has Dipankar Sarkar written books?+
Yes. He is the author of AI for Everyday Automation from Packt, published in 2026, and Nginx 1 Web Server Implementation Cookbook, published by Packt in 2011. He also publishes the open GenAI and Agentic AI Playbooks at What Generative AI.
Where is Dipankar Sarkar based?+
St Andrews, Scotland, United Kingdom. He works remotely with organisations and audiences internationally.