Founder & CEO β OM Enterprises Β· AI Research & Engineering Labs Senior ML Engineer Β· AI Researcher Β· 14+ years building production AI systems
π Portfolio: cataluna84.github.io β full project narrative, ARC-AGI work, Kaggle competitions, experience timeline
π Portfolio source: cataluna84/portfolio
Building ubiquitous, interactive intelligence that runs on multimodal data (audio, text, image, video) with high throughput and low latency.
My research spans Computer Vision, State Space Models (S4 / S5 / Mamba), Multimodal Models, Reinforcement Learning, Mechanistic Interpretability, and LLM compression β with applied experience deploying on-device (iOS), on edge (DeepStream SDK), and at scale (CUDA/Triton).
| π₯ ARC Prize 2024 | 24th / 1,427 teams β Silver Medal |
| π₯ ARC Prize 2025 | 87th / 1,455 teams β solo Bronze Medal |
| π ARC Prize 2026 | Active competitor |
| π₯ HF Flax/JAX Sprint | 3rd place β CLIP-RSICD (satellite images) |
| π Kaggle | Competitions Expert Β· rank ~1,532 / 202,876 Β· 1 Silver + 3 Bronze across 21+ competitions |
| π€ HuggingFace | 14 models published (NER, text generation, image generation) |
π¬ LID β Layer-Wise Multilingual Language ID β Layer-wise dynamics of multilingual language identification across 67 languages in compact foundation models. PyTorch Β· LoRA Β· Macro F1 0.97+
π§ ARC Prize 2026 (ARC-AGI-3) β Public lab notebook for the $850K ARC-AGI-3 competition. Agent zoo, FORGE, BFS, CNN, frame segmenter.
π΅ Codec Fine-tuning β Tiny Aya β Benchmark for fine-tuning neural speech codecs (Mimi, DualCodec, Kanade) on low-resource languages with 8 optimizers + W&B Bayesian sweeps.
π‘οΈ Crosslingual Emergent Misalignment β Cohere Labs research on how safety guardrails degrade across high- and low-resource languages.
π¨ Generative Deep Learning β VAE, GAN, and Diffusion implementations for image synthesis.
π Jamba β Hybrid TransformerβMamba β PyTorch implementation exploring SSM + attention for infinite-context LMs.
π World Models (Ha & Schmidhuber) β VAE + MDN-RNN + CMA-ES for RL in the OpenAI Car Racing environment.
β Full project list with metrics, write-ups, and demos: cataluna84.github.io
- Google DeepMind β
penzai(JAX neural network library) - TransformerLens β
TransformerLens(mechanistic interpretability) - Kyutai Labs β
moshi(open-source speech-text foundation model)
Deep Learning & Research
-7c3aed?style=flat){kind=icon}
ML Engineering & MLOps
-555555?style=flat){kind=icon}
Computer Vision & Edge
Languages & Tools
- Cohere Labs β Community Lead (Computer Vision sub-field) & Researcher
- TWiML (This Week in ML) β Contributor Lead, 5+ years moderating weekly GenAI meetups
- Lucknow AI Labs β Core organizer, Research Paper Team lead
- Machine Learning Tokyo (MLT) β Contributor (2019β2023)
- Yannic Kilcher Discord β Daily/weekly paper discussions
Conferences attended (remotely): ICLR Β· ICML Β· CVPR Β· NeurIPS
Open to Senior ML Engineer Β· Research Engineer Β· CV Engineer Β· Senior SDE roles, research collaborations, and advisory work.
- π§ mayankbhaskar007@gmail.com
- π cataluna84.github.io
- πΌ linkedin.com/in/cataluna84
- π kaggle.com/cataluna84
- π€ huggingface.co/cataluna84
- π x.com/cataluna84
- π Lucknow, Uttar Pradesh, India