"I want to understand intelligence — so I'm building it from scratch."
I'm a 2nd-year B.Tech IT student at DJSCE, Mumbai obsessed with the mechanics of deep learning. Not the API-call kind — the rewrite the transformer from scratch, read the original paper, understand every gradient kind.
My goal is to contribute research at the intersection of LLMs, Mech Interp and Computer Vision, with a long-term target of Masters in AI/ML.
parva = {
"currently" : "Sem 4 · B.Tech IT · DJSCE Mumbai",
"focus" : ["LLMs", "Vision Transformers", "Mamba", "NLP", "CV"],
"building" : "research profile from ground up",
"next" : "first publication + international research internship",
"motto" : "from first principles, always"
}
| Course | Provider | Status |
|---|---|---|
| Machine Learning Specialization | Andrew Ng — DeepLearning.AI + Stanford | Completed |
| Deep Learning Specialization | Andrew Ng — DeepLearning.AI | Completed |
| NLP Specialization | Andrew Ng — DeepLearning.AI | Completed |
| Intro to Large Language Models | NPTEL — IIT Delhi & IIT Bombay | Completed |
| Neural Networks for CV and NLP | NPTEL — IIT Guwahati | Completed |
| CS231N | Stanford Online (Youtube) | Completed |
| CS153 Frontier Systems | Stanford Online (Youtube) | In Progress |
| HuggingFace CV Course | HuggingFace | In Progress |
┌──────────────────────────────────────────────────────────────────┐
│ │
│ MECHANISTIC INTERP → TransformerLens, SAELens │
│ LANGUAGE MODELS → Pretraining, SFT, RLHF, scaling │
│ VISION TRANSFORMERS → ViT, Swin, dense prediction tasks │
│ MULTIMODAL AI → Vision-language alignment │
│ EFFICIENT INFERENCE → Quantization, distillation, pruning │
│ │
└──────────────────────────────────────────────────────────────────┘
