Amu Chandra - Founder, Builder, Linguist

About

Founder, builder, polyglot.

I'm a founder, product builder, polyglot and linguist. I have Master's degrees in linguistics, philology and phonetics from Oxford and in entrepreneurship from Cambridge, both with Distinction. I build AI products and multi-agent systems - connecting models to tools, data and real workflows so they can plan, act and get things done, not just answer questions. I also build with agents: coding agents and MCP-connected tooling are how my products get made. I advise others building the same way. My entrepreneurial experience spans technology, language and consumer goods.

I'm interested in learning and skill acquisition, with a particular focus on language learning. I speak nine languages. My approach is empirical: I synthesise existing research with practical experience to model how people acquire language.

Areas of Focus

Where I work & think

AI Systems

AI-native products and multi-agent systems - models that plan, act and use tools, connected to real data and workflows.

Product Development

Taking products from concept to launch - combining product strategy, rapid prototyping, AI-native development and iteration with real users.

Venture Building

Taking ideas from first principles to funded, operating companies.

Entrepreneurship

Full-cycle venture experience across technology, consumer and EdTech.

Consulting

Project work with founders and firms - AI product, multi-agent systems, and bringing agents into the way a team builds.

Language Science

Theoretical and applied work on language, learning and acquisition - from grammar and probability to models of how people learn languages effectively.

Ventures

As a Founder

Inana Labs

AI · Venture Studio

An AI-native product and venture studio building software, games, interactive experiences and new production systems with artificial intelligence at the core.

We combine coding agents, multimodal models and agentic workflows to move rapidly from idea to working product - developing new ventures, digital products, games and AI-native production systems.

AI Systems Agentic Workflows Product Development Interactive Worlds

Letera AI

EdTech · AI

An AI-powered language learning platform built on the comprehensible input hypothesis and personalised multimodal learning. Letera took an empirical, research-driven approach to language acquisition.

Finalists at Oxbridge AI. Supported by Accelerate Cambridge and UCL EdTech Labs. Partnerships at the University of Cambridge and the University of Oxford.

AI/ML EdTech Language Learning SaaS

Ishara Organic

Consumer Goods · FMCG

An ethical consumer goods brand bringing organic and sustainable products to premium UK and international retail. Stocked at Ocado, Selfridges, Fortnum & Mason, and TJX.

Featured by BBC and Vogue. Built from first principles - sourcing, brand, distribution - demonstrating that ethical production and commercial success are not in conflict.

FMCG Ethical Consumer Sustainable Sourcing Premium Retail

Languages

Polyglot. Linguist.
Language Researcher.

I hold a Master's degree in Linguistics from the University of Oxford, where my research focused on syntax, psycholinguistics and the structure of language. My interests in language centre on human language acquisition and the cognitive mechanisms that underpin how we learn and internalise new languages.

I speak nine languages, and that experience is inseparable from my research. My approach follows usage-based, input-driven models of acquisition - the view that language is learned through meaningful exposure at the right level of difficulty, with explicit instruction in a supporting rather than a leading role.

The success of deep learning and large language models offers a striking validation of this view. LLMs acquire language not through explicit rules, but through massive exposure to high-quality input - converging on probabilistic representations of grammar, meaning and usage. This mirrors what usage-based and statistical-learning accounts have long argued about human learners: that quality input, at the right level of comprehensibility, is the primary driver of acquisition. These insights inform both my research and the language tools I build.

Academic

Academic Background

University of Cambridge

MSt Entrepreneurship

Distinction

Award-winning dissertation Research on entrepreneurial cities and global innovation hubs. Scored the highest mark in the history of the programme. Explored how urban policy, talent density, and capital concentration interact to produce startup ecosystems.

University of Oxford

MSt Linguistics, Philology & Phonetics

Distinction

Language acquisition & generative theory Focused on syntax, psycholinguistics and the structure of language, with particular attention to the limits of Chomskyan generative grammar. Research laid the groundwork for later work on comprehensible input and probabilistic language models.

King's College London

BA English Literature

Language, literature & critical theory Undergraduate study in English Literature, developing a foundation in textual analysis, critical theory and the history of language - informing later postgraduate work in linguistics and language acquisition.

Writing

Selected Papers

Entrepreneurship · Urban Economics

Entrepreneurial cities in the 21st century: How cities can generate entrepreneurship, attract talent and stimulate growth in the digital age

Amu Chandra - University of Cambridge

Award-winning dissertation submitted for the MSt Entrepreneurship at the University of Cambridge. Explores how urban policy, talent density, and capital concentration interact to produce startup ecosystems - and what distinguishes cities that sustain innovation from those that do not. Scored the highest mark in the history of the programme.

Download PDF Preview

Linguistics · Language Acquisition

From Competence to Usage: Distributional Semantics, Connectionism, and the Case for a Usage-Based Paradigm in Language Acquisition

Ezio Xie & Amu Chandra · Working paper, September 2026

Working Paper

A theoretical synthesis arguing that usage-based and constructionist linguistics, connectionist cognitive science, and distributional language models converge on an account of linguistic knowledge as emerging from statistical regularities in use. The paper considers implications for second-language acquisition, including frequency-driven learning, input, and what contemporary language models can - and cannot - tell us about human acquisition.