Lossfunk: the AI lab for independent researchers
Hi! I'm Paras, founder of Lossfunk. I'm a newly-minted AI researcher [my Google Scholar profile] and a builder [my GitHub profile]. Previously, I founded Wingify, bootstrapped it to $50mn ARR and then exited to a private equity group.
Now, with Lossfunk, I'm building an AI research lab for curious people and independent researchers.
Programs
- Initialize (for independent researchers): conduct AI research part time remotely. Join our invite-only community of researchers, get compute/credits for your project
- Residency (for curious minds): a 6 week, on-site, curiosity-led exploration into AI and related topics
- Research Internship (for students): internship for those who are exploring a career in AI research (on their way to a PhD, or independent research)
Lab's focus
We're interested in foundational questions in AI and beyond:
- Nature of reality: is there a mathematical multiverse? can the anthropic principle prove it?
- Self-actualizing patterns: How does "self" come into existence? what does agency look like in abstract spaces?
- Intelligence in brain: how to model brain dynamics? what can we learn from "odd" phenomena in brain (like illusions, dreams, etc)?
- Consciousness: what is consciousness? what can we learn from unconventional minds (such as an ant colony)?
- Learning representations: is manifold hypothesis true? how do deep networks learn representations?
- AI (general intelligence): what is artificial general intelligence, and how to create it?
- AI (safety): is orthogonality thesis false? what can homeostatis teach us about AI safety?
- AI (deep learning): how to solve continual learning? how to achieve sample efficiency?
- AI (agents): how to create agents that can do long-horizon tasks without getting derailed?
- AI accelerated science: how to accelerate science with AI? what tools and systems are needed?
- Novelty creation by AI: how to mathematically model novelty?
- Philosophy: can we use AI to answer what is beauty, ethics, and other philosophical questions?
Check out this spreadsheet for the full list of research questions Lossfunk is interested in.
Projects underway
We have 25+ projects underway. Here are some of them:
- World models for games: Can an AI learn to play a game (with minimal data) purely from its imagination?
- AI scientist: How science will be done if we (re)invented it today with AIs
- AI engineer: Can an AI system solve hard AI/ML engineering problems autonomously?
- Meta RL: can we discover sample efficient RL algorithms via evolutionary algorithms?
- Automatic CUDA optimization: Can we get AI to improve GPU kernels, speeding up AI training and inference?
- Language - culture entanglement: Do LLMs reason differently in different languages?
- Reasoning in vision models: Can we get strong reasoning in vision models?
- AI forecaster: Can we get AI to predict how future events will turn out?
- AI x Ethics: How can we use AI to explore human ethical systems and norms?
- India@ML: papers by Indian teams accepted in top AI/ML conferences
Publications and articles
Talks
We regularly host talks on AI and related topics at our office in Bangalore. You can see our past talks on our youtube channel.
To get notified about upcoming talks, join our Whatsapp channel.
Links
-[--->+<]>---.+[----->+++<]>.----.+++.+[--->+<]>++.-[---->+<]>++.---[->++++<]>.-----.[--->+<]>-----.++++++[->++<]>.+[--->+<]>.++++..-------------.-[--->+<]>--.-------.---.[->+++<]>--.
Stay curious :)
.
.
.
.
.
.
Still there? Here's a wall of testimonials
^^^ World #25 player in chess
