High-quality data for pre- and post-training
Curation, filtering, and synthesis pipelines for Indic and Global South text, code, math, and speech — built for both pre-training and alignment stages.
// PROJECT EKΛ
Backed by the IndiaAI Mission
Sovereign models for reasoning & multilingual AI
Project EKΛ is Soket's flagship research program — advancing frontier math, code, and reasoning models in parallel with sovereign multilingual AI for Indian and Global South languages.
// WHAT WE'RE BUILDING
Foundation models built for Bharat & the Global South
Project EKΛ is Soket’s boldest vision — building AI for a billion, from the heart of India. Our mission is to create world-class models that master math, code, and reasoning, while speaking the languages of Bharat and the Global South. We believe talent and ambition from India can shape the very frontier of AI. We work at the edge of research in architecture, large-scale training, and language resources — reimagining what’s possible for low-resource and diverse languages. Join us at the vanguard — and help put India at the center of global AI innovation. Project EKΛ is dedicated to producing fundamental advances in both the science and engineering of AI.
We open-source where we can, train on sovereign compute, and publish research that advances Indic NLP, systems for ML, and efficient inference.
1536
NVIDIA H100 GPUs
Sovereign compute via the IndiaAI Mission
25T
Tokens curated
High-quality pre-training corpus to date
120B+
Parameters (Sparse MoE)
Frontier-scale architecture in training
60+
Languages
Indian, Global South, and programming languages
// CORE CAPABILITIES
Two tracks: technical reasoning & multilingual AI
Math
Advanced mathematical reasoning, proofs, and symbolic computation — models tuned for rigorous step-by-step logic.
Code
Multi-language code generation, debugging, and optimization across 20+ programming languages.
Reasoning
Logical deduction, complex analysis, and long-horizon problem solving for high-stakes workflows.
Multilingual
22 Indian languages, 20+ Global South languages, and English — a dedicated vertical for sovereign text and speech, with curated data and tokenization built for linguistic diversity.
// LANGUAGE COVERAGE
60+ languages — not an afterthought
Most frontier labs optimize for English. EKΛ runs two parallel verticals: frontier math, code, and reasoning models for rigorous technical work — alongside sovereign multilingual modeling for Indian and Global South languages, with efficient tokenization and curated corpora per script for regional text, speech, and domain-specific use cases.
22 Indian languages
Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Odia, Punjabi, Assamese, Urdu, Sanskrit, and more
20+ Global South languages
Arabic, Indonesian, Thai, Vietnamese, Burmese, Kazakh, Portuguese, Spanish, and more
20+ programming languages
Python, Rust, Go, TypeScript, C++, Java, SQL, Julia, and more
// RESEARCH DIRECTIONS
Problems we're actively working on
If you care about data systems, training at scale, tokenizers, post-training, or ethical AI — these are the threads where your work ships into a national-scale model, not a side project.
Efficient model architecture for language variance
Sparse MoE and routing strategies that handle extreme token-efficiency and morphological diversity across scripts and domains.
Efficient tokenization
Building one of the most token-efficient vocabularies for Indian and Global South languages — minimizing bytes-per-token for Indic scripts and enabling longer context at lower compute cost.
Post-training methods
SFT and preference optimization for math, code, and reasoning models; separate alignment pipelines for multilingual foundations and regulated-sector workflows.
Efficient inference & algorithmic design
Kernel fusion, speculative decoding, quantization, and systems co-design from training through production serving.
Sustainable & efficient AI research
Optimal power and water usage for large-scale training — measuring and minimizing the environmental cost of frontier runs.
AI for critical sectors
Applied research for defence, cybersecurity, military and civilian intelligence, finance, banking, and information technology — with auditable, on-premise deployment paths.
Ethical AI
Safety, alignment, bias mitigation, and responsible deployment — embedding ethical constraints into data curation, training, evaluation, and release for sovereign and high-stakes use cases.
// RELATED RELEASES
EKA tokenizer
Among the most token-efficient vocabularies for Indian and Global South languages — built in-house to cut sequence length and training cost versus mainstream open tokenizers.
Help us train India's frontier models
We're hiring researchers and engineers across data, training, inference, and applied ML. If you want hard systems problems at sovereign scale — we'd like to hear from you.