I'm a pre-final year ECE undergrad at SGGS Nanded who got pulled into the AI rabbit hole and never came back.
My path started with circuits and signals — but somewhere between understanding how neurons fire and how transformers attend, I realized the most interesting engineering happening right now is at the intersection of language, reasoning, and autonomous systems. So that's where I went.
I don't collect projects — I build things I want to understand deeply.
Not fine-tuning, not LoRA on someone else's weights. A full GPT architecture (6 layers, 6 heads, 384-dim) trained on TinyStories from scratch — custom tokenization, mmap pipelines, AMP mixed precision, warmup + cosine LR scheduling. Built to understand what's actually happening inside the attention mechanism.
A live system where 4 specialized agents (Ordering, Safety, Forecast, Procurement) collaborate autonomously — prescription parsing, inventory reasoning, order generation across 4 languages. ChromaDB for semantic retrieval, Langfuse for full observability of every LLM reasoning step. Not a chatbot wrapper — a real agentic architecture.
Desktop application for detecting, analyzing, and visualizing RF signals in real time using HackRF One. Clean four-layer architecture: UI → EngineController → Engine pipeline → IQ hardware. Physical-layer first — no protocol decoding, no black-box classification. Everything is deterministic and traceable. Dual modes: Inspector and Scanner.
ML / AI
LLM & Agentic
Backend & Infra
Databases
|
|
|
AI/ML Research · LLM Engineering · Agentic AI Engineering
I want to work on systems that reason, not just autocomplete. If you're building something in that space, I'm interested.
▸ Research Interests
- LLM scaling and emergent capabilities
- Efficient inference and model compression (LoRA, PEFT, INT4/INT8 quantization)
- Multi-agent coordination and tool-augmented reasoning
- Retrieval-Augmented Generation (RAG) for long-context tasks
- Transformer pre-training dynamics
▸ Certifications
- Quantization Fundamentals — Hugging Face
- MCP: Build Rich-Context AI Apps — Anthropic
- AI Fluency: Framework & Foundations — Anthropic