Ryta Zasiekina

We believe in the trinity: model, inference stack, hardware. Companies that focus on a single component of this trinity lack sovereignty and are constrained by the architectural choices made by others. Most labs treat on-device models as scaled-down versions of their cloud-focused cousins. But LLM architectures that evolved for the cloud are not well-suited to on-device setups. Cloud LLMs operate in the arithmetic-bound regime. Mainstream architectures aim to maximise total token throughput by reducing the amount of computation performed per request, and they treat device memory as an unlimited resource. But for on-device deployment, memory is the main bottleneck, both in terms of throughput and the size of the resident set. When designing our on-device architecture, we focus on three core objectives: increasing the arithmetic intensity of the decoding stage, reducing the size of the resident set, and maximally utilising the GPU neural accelerators. This leads us to models that differ from traditional autoregressive transformers in a number of meaningful ways.

The Hands In mission is simple: to eliminate payment hurdles like 𝗱𝗲𝗰𝗹𝗶𝗻𝗲𝘀 𝗱𝘂𝗲 𝘁𝗼 𝗶𝗻𝘀𝘂𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝗳𝘂𝗻𝗱𝘀, drive 𝗶𝗻𝗰𝗿𝗲𝗺𝗲𝗻𝘁𝗮𝗹 𝗿𝗲𝘃𝗲𝗻𝘂𝗲 for our business partners, and create smoother transactions that improve customer satisfaction. 𝗧𝗵𝗲 𝗣𝗮𝘆𝗺𝗲𝗻𝘁 𝗣𝗿𝗼𝗯𝗹𝗲𝗺 𝗮𝗻𝗱 𝗧𝗵𝗲 𝗛𝗮𝗻𝗱𝘀 𝗜𝗻 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻 Travel transactions declining due to insufficient funds make up to on average 5% of all failing payments, contributing to millions lost in revenue at the checkout every month. Additionally, 59% of travelers travel together in groups every year but cannot split the payment. And thats exactly what Hands In wants to solve... We’ve designed three groundbreaking features to tackle these issues head-on: • 𝗠𝘂𝗹𝘁𝗶-𝗖𝗮𝗿𝗱 𝗣𝗮𝘆𝗺𝗲𝗻𝘁𝘀: Customers can easily split costs across multiple cards, allocating amounts and authorizing each payment in seconds. • 𝗚𝗿𝗼𝘂𝗽 𝗣𝗮𝘆𝗺𝗲𝗻𝘁𝘀: Whether it’s friends booking a trip or colleagues organizing an event, peers can join the same transaction and pay their share—no more chasing people down for reimbursement. • 𝗗𝗲𝗰𝗹𝗶𝗻𝗲 𝗥𝗲𝗰𝗼𝘃𝗲𝗿𝘆: Customers can split payments across multiple cards if their original payment method is declined due to insufficient funds, allowing two card transaction retries, versus just one. 𝗣𝗿𝗼𝘃𝗲𝗻 𝗦𝘂𝗰𝗰𝗲𝘀𝘀: Generating over $30M in incremental sales for clients in 2025. 𝗢𝘂𝗿 𝗣𝗮𝗿𝘁𝗻𝗲𝗿𝘀𝗵𝗶𝗽𝘀 • Air Europa • AsiaPay • Budgy Smuggler • Arab Financial Services • Sports Events 365 • Repayd • Cell Point Digital • BR-DGE • YUNO • DEUNA • BridgerPay Our API connects to up to 𝟰𝟬 𝗣𝗦𝗣𝘀, including major names like 𝗪𝗼𝗿𝗹𝗱𝗽𝗮𝘆, 𝗘𝗹𝗮𝘃𝗼𝗻, and 𝗔𝗱𝘆𝗲𝗻. 𝗔𝘄𝗮𝗿𝗱𝘀, 𝗥𝗲𝗰𝗼𝗴𝗻𝗶𝘁𝗶𝗼𝗻𝘀, 𝗮𝗻𝗱 𝗙𝗲𝗮𝘁𝘂𝗿𝗲𝘀: https://www.handsin.com/about-us Join the payment revolution with Hands In, and let’s shape the future of payments together!

We the devs, are passionate about VR, games and creation. On a mission to bring about the moment when all people can have access to indistinguishable virtual reality experience <3

Teramot | The Connectivity Layer for AI Agents Los Agentes de IA están transformando cómo las empresas operan. Pero para que realmente funcionen, necesitan acceso a los datos reales del negocio. Ahí entra Teramot. Desarrollamos una plataforma que permite construir, conectar y escalar Agentes de Última Milla—aquellos que piensan, razonan y actúan directamente sobre los sistemas reales de las compañías. 🔹 10X más velocidad 🔹 10X menos costo 🔹 Cero dependencia de equipos técnicos internos Con Teramot, los agentes dejan de ser una promesa y se convierten en producción.

We believe in the trinity: model, inference stack, hardware. Companies that focus on a single component of this trinity lack sovereignty and are constrained by the architectural choices made by others. Most labs treat on-device models as scaled-down versions of their cloud-focused cousins. But LLM architectures that evolved for the cloud are not well-suited to on-device setups. Cloud LLMs operate in the arithmetic-bound regime. Mainstream architectures aim to maximise total token throughput by reducing the amount of computation performed per request, and they treat device memory as an unlimited resource. But for on-device deployment, memory is the main bottleneck, both in terms of throughput and the size of the resident set. When designing our on-device architecture, we focus on three core objectives: increasing the arithmetic intensity of the decoding stage, reducing the size of the resident set, and maximally utilising the GPU neural accelerators. This leads us to models that differ from traditional autoregressive transformers in a number of meaningful ways.

The Hands In mission is simple: to eliminate payment hurdles like 𝗱𝗲𝗰𝗹𝗶𝗻𝗲𝘀 𝗱𝘂𝗲 𝘁𝗼 𝗶𝗻𝘀𝘂𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝗳𝘂𝗻𝗱𝘀, drive 𝗶𝗻𝗰𝗿𝗲𝗺𝗲𝗻𝘁𝗮𝗹 𝗿𝗲𝘃𝗲𝗻𝘂𝗲 for our business partners, and create smoother transactions that improve customer satisfaction. 𝗧𝗵𝗲 𝗣𝗮𝘆𝗺𝗲𝗻𝘁 𝗣𝗿𝗼𝗯𝗹𝗲𝗺 𝗮𝗻𝗱 𝗧𝗵𝗲 𝗛𝗮𝗻𝗱𝘀 𝗜𝗻 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻 Travel transactions declining due to insufficient funds make up to on average 5% of all failing payments, contributing to millions lost in revenue at the checkout every month. Additionally, 59% of travelers travel together in groups every year but cannot split the payment. And thats exactly what Hands In wants to solve... We’ve designed three groundbreaking features to tackle these issues head-on: • 𝗠𝘂𝗹𝘁𝗶-𝗖𝗮𝗿𝗱 𝗣𝗮𝘆𝗺𝗲𝗻𝘁𝘀: Customers can easily split costs across multiple cards, allocating amounts and authorizing each payment in seconds. • 𝗚𝗿𝗼𝘂𝗽 𝗣𝗮𝘆𝗺𝗲𝗻𝘁𝘀: Whether it’s friends booking a trip or colleagues organizing an event, peers can join the same transaction and pay their share—no more chasing people down for reimbursement. • 𝗗𝗲𝗰𝗹𝗶𝗻𝗲 𝗥𝗲𝗰𝗼𝘃𝗲𝗿𝘆: Customers can split payments across multiple cards if their original payment method is declined due to insufficient funds, allowing two card transaction retries, versus just one. 𝗣𝗿𝗼𝘃𝗲𝗻 𝗦𝘂𝗰𝗰𝗲𝘀𝘀: Generating over $30M in incremental sales for clients in 2025. 𝗢𝘂𝗿 𝗣𝗮𝗿𝘁𝗻𝗲𝗿𝘀𝗵𝗶𝗽𝘀 • Air Europa • AsiaPay • Budgy Smuggler • Arab Financial Services • Sports Events 365 • Repayd • Cell Point Digital • BR-DGE • YUNO • DEUNA • BridgerPay Our API connects to up to 𝟰𝟬 𝗣𝗦𝗣𝘀, including major names like 𝗪𝗼𝗿𝗹𝗱𝗽𝗮𝘆, 𝗘𝗹𝗮𝘃𝗼𝗻, and 𝗔𝗱𝘆𝗲𝗻. 𝗔𝘄𝗮𝗿𝗱𝘀, 𝗥𝗲𝗰𝗼𝗴𝗻𝗶𝘁𝗶𝗼𝗻𝘀, 𝗮𝗻𝗱 𝗙𝗲𝗮𝘁𝘂𝗿𝗲𝘀: https://www.handsin.com/about-us Join the payment revolution with Hands In, and let’s shape the future of payments together!

We the devs, are passionate about VR, games and creation. On a mission to bring about the moment when all people can have access to indistinguishable virtual reality experience <3

Teramot | The Connectivity Layer for AI Agents Los Agentes de IA están transformando cómo las empresas operan. Pero para que realmente funcionen, necesitan acceso a los datos reales del negocio. Ahí entra Teramot. Desarrollamos una plataforma que permite construir, conectar y escalar Agentes de Última Milla—aquellos que piensan, razonan y actúan directamente sobre los sistemas reales de las compañías. 🔹 10X más velocidad 🔹 10X menos costo 🔹 Cero dependencia de equipos técnicos internos Con Teramot, los agentes dejan de ser una promesa y se convierten en producción.

The game-changing expertise in payments and banking at a place where every voice is valued and every view is respected.