I am a founder and product builder applying AI to solve tangible human problems. I believe the most powerful innovations come from identifying a real-world frustration and building a solution from the ground up. I put this into practice by founding LuarAI, where I co-founded and led the technical execution for four distinct software products in three months. My approach combines rapid prototyping with a deep understanding of AI systems, from large language models to my own research in reinforcement learning. I am seeking my next challenge in a technical product or founding role where I can bridge the gap between an ambitious vision and real-world execution.

Juan Sanchez

I am a founder and product builder applying AI to solve tangible human problems. I believe the most powerful innovations come from identifying a real-world frustration and building a solution from the ground up. I put this into practice by founding LuarAI, where I co-founded and led the technical execution for four distinct software products in three months. My approach combines rapid prototyping with a deep understanding of AI systems, from large language models to my own research in reinforcement learning. I am seeking my next challenge in a technical product or founding role where I can bridge the gap between an ambitious vision and real-world execution.

Available to hire

I am a founder and product builder applying AI to solve tangible human problems. I believe the most powerful innovations come from identifying a real-world frustration and building a solution from the ground up. I put this into practice by founding LuarAI, where I co-founded and led the technical execution for four distinct software products in three months.

My approach combines rapid prototyping with a deep understanding of AI systems, from large language models to my own research in reinforcement learning. I am seeking my next challenge in a technical product or founding role where I can bridge the gap between an ambitious vision and real-world execution.

See more

Experience Level

Expert
Expert
Expert
Expert
Expert
Expert
Expert
Expert
Intermediate
Intermediate
Intermediate
Beginner
See more

Language

Spanish; Castilian
Fluent
English
Advanced
Portuguese
Intermediate
Chinese
Beginner
German
Beginner

Work Experience

Co-Founder & Head of Product/Technology at LuarAI
June 1, 2025 - Present
Conceived, co-designed, and led technical development of four distinct AI-powered applications in three months, taking each from initial concept to live MVP. Launched products include SAN FAN SON (LLM-powered language learning tutor), Zenota (hierarchical, canvas-based knowledge management tool), Divo (no-code visual web editor for responsive layouts), and Picky (AI-driven tool for instant restaurant menu digitization). Architected and built full-stack systems leveraging GPT API, Gemini API, modern web frameworks (React, Node.js), and cloud deployment platforms. Partnered with a co-founder to lead product strategy, translating user needs and market analysis into functional specifications and intuitive UI/UX designs. Drove entire product life cycle from ideation and user problem validation to go-to-market planning and initial user feedback collection.
Architect Lead at Parameta
April 30, 2024 - August 30, 2025
Architected and led implementation of enterprise Business Spend Management (Coupa) solutions for multinational clients across LATAM. Developed Python-based automation tools that reduced architects' manual task time from 40% to under 20%, significantly boosting team productivity. Managed a high-stakes project delivering a full Coupa implementation in a record two months.
Life Cycle Program Manager at Nub 7/8
July 31, 2023 - August 30, 2025
Managed full program life cycle for SaaS clients using the Cisco Smart Net Total Care portal, focusing on onboarding, deployment, and cross-cultural communication. Created Python scripts within Google Colab that automated operational reporting, accelerating key processes by over 600%.

Education

Bachelor of International Business Administration at University of La Sabana
September 1, 2025 - September 2, 2025
Google Data Analytics Professional Certificate at Coursera
November 1, 2022 - November 30, 2022

Qualifications

Google Data Analytics Professional Certificate
November 1, 2022 - November 30, 2022

Industry Experience

Software & Internet, Education, Professional Services, Computers & Electronics
    uniE608 Loss as a Reward LaaR
    📉 Loss as a Reward 🎁 In reinforcement learning, designing the “right” reward function can be more challenging than building the model itself. That’s why we explored LaaR (Loss as a Reward), an approach where we forgo traditional RL reward shaping and instead let the supervised training loss itself guide the agent. Why is this exciting? • Automatic Reward Signal: By using the classification loss directly, we sidestep the need for hand-crafted reward functions. The agent learns to minimize loss naturally, just like a supervised model, but does so through a sequence of actions that zooms in on the pertinent parts of the input. • Partial Observability & Efficiency: We’re flipping the paradigm of analyzing all pixels at once. Instead, our agent takes a limited “window” glimpse at the image, much like how humans momentarily focus on different parts of a scene. This is hypothetically more computationally efficient and more biologically plausible. We limit the agent to 10 glimpses, but interestingly, it already recognizes the digit by the third glimpse. • Bridging RL and Supervised Learning: Rather than treating reinforcement learning and supervised learning as separate paradigms, this approach intertwines them. The RL agent’s policy improvement is driven by the same metric used to train conventional neural networks. Under the Hood Decision Transformer: At the core is a Decision Transformer (DT) that handles sequences of states and actions just like language models handle tokens. We feed it the agent’s past observations, actions, and (inverted) loss-based reward. Curriculum Learning: We take a cue from how students learn, building a solid foundation before tackling advanced material. Initially, the agent sees “easier” digits; tackling too-challenging examples from the get-go could result in confusion and poor learning. Dynamic Window & Zoom: The environment allows the agent to move a small window across the image and zoom in or out, effectively deciding where to look next. 86.6% Accuracy: This simple yet effective architecture, treating each classification step as an RL step, achieves an 86.6% accuracy on MNIST test. Implications • Generalizable Reward: Wherever there’s a differentiable training objective (like cross-entropy loss), we can convert it into a reinforcement signal. • Guided Exploration: By tying the reward to the loss, the agent explores actions that directly reduce errors, requiring fewer hyperparameters or custom reward hacking. • Scalability: The idea can be extended to larger images, multi-class tasks, or problems where typical RL reward signals are hard to define. AIArchitecture Architecture ReinforcementLearning RL
    uniE621 Luarai
    luarai.com My company website/landing page website startup landingpage
    uniE608 Zenota
    zenota.luarai.com A visual-first note-taking app that combines a spatial canvas with structured hierarchies. Zenota features an AI assistant, "Structo," to automate note organization, making it ideal for visual thinkers who need clarity and structure. NoteTaking AI Productivity
    uniE621 Sanfanson
    sanfanson.luarai.com A language-learning game that uses interactive minigames and an AI-powered feedback system to make learning new languages engaging and effective. Features include leaderboards, progress tracking, and dynamic content generation. Gamification AI EdTech
    uniE621 Picky
    An interactive and intelligent digital menu for restaurants. Picky allows users to filter dishes by ingredients, allergens, and dietary needs. It also features an AI-powered admin view ("Pickyto") to help restaurant owners digitalize their menus effortlessly. FoodTech AI SaaS
    uniE621 Koby
    koby.luarai.com An insightful web app that lets you upload, visualize, and share your Kobo e-reader highlights. Koby transforms your reading data into beautiful charts and shareable insights, giving you a new perspective on your literary journey. https://www.twine.net/signin E-Reader DataViz Firebase
    uniE621 Divo
    divo.luarai.com A whimsical, visual-first web editor that lets you vibe your site into existence. Drag, drop, and tweak your way to a beautiful, responsive webpage, then export clean HTML with a single click. Built for designers, creators, and anyone who thinks coding should be more like playing. VisualEditor NoCode WebDesign