# Our Team — Gen.new | AI Research & Systems Engineering

> Meet Nikita Voloboev and Arnesh Banerjee, the systems engineering and machine learning team building Gen.new’s unified AI creative workspace.

[Read this page on Gen.new](https://gen.new/team)

We’re building Gen.new, a unified AI creative workspace where people and AI agents work together across text, images, video, audio and 3D. Our team brings systems engineering, open-source software and machine learning research into one creative workflow.

## Nikita Voloboev

Founder & AI Architect

@nikivdev

Systems engineer and open-source creator focused on AI agents, knowledge tools and responsive software.

Nikita leads Gen.new’s product and AI architecture. His work brings systems engineering and human–agent collaboration together, with a focus on fast interactions and tools that make complex workflows easier to use.

He created Learn Anything, an open-source knowledge graph for discovering learning resources, and maintains a public digital knowledge base alongside developer tools in TypeScript, Rust, Go and C++.

### Selected open-source work

- [Learn Anything](https://learn-anything.xyz/)
  
  An open-source learning platform that connects topics with public learning resources.
- [Gen open source](https://github.com/gendotnew/ai)
  
  Public project work in the Gen ecosystem, continuing a history of open knowledge and creative tools.
- [Go tools & libraries](https://github.com/nikivdev/go)
  
  A collection of command-line tools and Go libraries for developer workflows.

[GitHub](https://github.com/nikivdev)

[Personal site & wiki](https://nikiv.dev/)

[X / Twitter](https://x.com/nikivdev)

[LinkedIn](https://www.linkedin.com/in/nikitavoloboev/)

## Arnesh Banerjee

Machine Learning Engineer

@ArneshBanerjee

Machine learning and systems engineer working on deep learning, large language models, AI safety and low-level performance.

Arnesh works across model training, reinforcement learning and efficient inference. His research experience includes IIT Kharagpur, Jadavpur University and the New Jersey Institute of Technology, spanning computer vision and LLM alignment.

His open-source work includes GPT-2 inference in ARM64 assembly and a rebuilt Kimi K3 training pipeline used to train a 1.27-billion-parameter test model. He also contributes to projects including CPython, Rust, Zed and OpenCV.

### Selected open-source work

- [GPT-2 in ARM64 assembly](https://github.com/ArneshBanerjee/gpt2.s)
  
  GPT-2 small, with 124 million parameters, implemented in hand-written AArch64 assembly without C or libc dependencies.
- [Kimi K3 for All](https://github.com/ArneshBanerjee/Kimi-K3-for-All)
  
  An independently rebuilt training pipeline with a 1.27-billion-parameter test model.

### Research & preprints

- [Certifiable Safe RLHF: Semantic Grounding and Fixed Penalty Constraint Optimization for Safer LLM Alignment](https://arxiv.org/abs/2510.03520)
  
  arXiv preprint
- [Failure Modes of Large Language Models on Research-Level Mathematics: A Taxonomy and an Empirical Characterisation](https://arxiv.org/abs/2606.24902)
  
  arXiv preprint

[GitHub](https://github.com/ArneshBanerjee)

[Personal site & blog](https://arneshbanerjee.dev/)

[Google Scholar](https://scholar.google.com/citations?hl=en&user=BzV2n8cAAAAJ)

[ORCID](https://orcid.org/0009-0007-1818-8477)

[X / Twitter](https://x.com/Arnesh_24)

[LinkedIn](https://www.linkedin.com/in/arnesh24/)

## One workspace. More room to create.

Gen connects AI models through Chat and the [Studio canvas](https://gen.new/studio). Start with a conversation, build [AI product photos](https://gen.new/ai-product-photography), turn an [image into a video](https://gen.new/image-to-video), or create [3D assets from a reference image](https://gen.new/image-to-3d).

[Try Gen free](https://gen.new/#create)
