AI Research Collective

We follow the questions; the answers become products.

A group of researchers pursuing the hardest problems we can find — and turning what we learn into products and partnerships.

Research is the origin. Products and partnerships are what it produces.

We don't start from a roadmap. We start from the questions we can't stop thinking about — and follow them until they turn into something real.

How we work

A collective, not an agency.

We're a group of people who share a set of questions and pursue them together. Everything the lab produces traces back to that.

The questions

Open problems in AI worth years of attention, pursued out of genuine interest rather than a roadmap.

The products

When the work produces something worth shipping, it becomes a product — like Vectora, our research platform.

The partnerships

When a client's problem lines up with what we're already exploring, we take it on as a partnership.

Research

The questions we're following.

Our work spans a few directions. Each one is a thread we're actively pulling.

LLM efficiency & cost optimization

Making large models cheaper to run without giving up quality.

Agentic systems

Systems that plan, use tools, and carry out multi-step work reliably.

Retrieval & research tooling

Better ways to find, read, and organize a field's literature.

Applied ML & data science

Turning messy real-world data into models that hold up in production.

LLM security

Defending models against prompt injection, data leakage, and training-data poisoning.

Privacy-preserving AI

Learning from sensitive data without centralizing it, through federated learning.

Neural topology

The intersection of algebraic topology, neuroscience, and AI — how network shape drives computation.

Alignment & hallucination reduction

Cutting false outputs with constitutional distillation, self-rewarding methods, and rigorous data curation.

Our first product In development

Vectora

A research platform for people who read papers for a living. Discover work through meaning, not keywords; ask an AI about any paper; and keep everything in a versioned research workspace you can share with collaborators.

It's the clearest example of our research becoming a product. Early access is coming.

Join the early access list
Services

Where our research meets your problem.

When a project aligns with what we're already exploring, we take it on. These are the areas we work in.

AI & Machine Learning

  • AI Integration
  • Agentic AI Development
  • LLM Cost Optimization
  • Machine Learning / Deep Learning
  • Data Science

Engineering & Infrastructure

  • Backend Development
  • Full-Stack Development
  • Data Engineering
  • DevOps & Cloud

We take on a limited number of engagements — the ones where the work teaches us something and moves our own research forward.

The collective

The people following the questions.

A small group with overlapping obsessions.

Hari Prasad

Hari Prasad

CEO

Ananthu S Kuttattu

Ananthu S Kuttattu

CTO

Ananthakrishnan Harikumar

Ananthakrishnan Harikumar

AI / ML Engineer

Gokul Gopan

Gokul Gopan

LLM Engineer

Sreenu Mappompilly Raj

Sreenu Mappompilly Raj

Full-Stack Engineer

Contact

Have a question that fits ours?

Tell us what you're working on. The best work tends to start where your questions meet ours.