โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ โ
โ ayush@void:~$ whoami โ
โ โ
โ NAME Ayush Kumar โ
โ ROLE AI Engineer โ
โ EDUCATION B.Tech AI & Data Science, 2027 โ
โ โ
โ FOCUS ML Engineering ยท LLM Engineering โ
โ Agentic AI ยท MLOps โ
โ โ
โ STACK Python ยท PyTorch ยท TensorFlow โ
โ LangGraph ยท FastAPI ยท Docker โ
โ AWS ยท MLflow ยท DVC โ
โ โ
โ STATUS training models. building agents. โ
โ shipping things that actually work. โ
โ โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
projects:
- name: GPT From Scratch
tech: PyTorch
note: decoder-only transformer,
first principles
- name: Agentic Research Assistant
tech: LangGraph ยท RAG
note: multi-agent autonomous
research pipeline
- name: Brain Tumor MRI Classifier
tech: TensorFlow ยท FastAPI
note: production medical imaging
classification system
- name: Production RAG Systems
tech: LangChain ยท FAISS
note: scalable retrieval-augmented
generation at scale |
learning:
- LLM Internals
# attention, KV caches, quantization
- Multi-Agent Systems
# orchestration, tool calling, memory
- Distributed Training
# FSDP, DeepSpeed, gradient checkpointing
reading:
- "Attention Is All You Need"
- "Scaling Laws for Neural LMs"
- "ReAct: Reasoning + Acting"
philosophy: >
if it doesn't solve a real problem,
delete it.
the best model is the one
that's actually deployed. |
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
midnight. rain. code. deploy. repeat.
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
|
GPT FROM SCRATCH Decoder-only transformer built from first principles.
|
AGENTIC RESEARCH ASSISTANT Multi-agent system for autonomous research workflows.
|
|
BRAIN TUMOR MRI CLASSIFIER Production-grade medical image classification system.
|
PRODUCTION RAG SYSTEMS Scalable retrieval-augmented generation pipeline.
|
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
"I'd rather be dry, but at least I'm alive." โโ the nbhd
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ โ
โ ayush@void:~$ cat /etc/motd โ
โ โ
โ โ ship fast. ship often. ship things that matter. โ
โ โ read papers. write code. repeat. โ
โ โ the best model is the one that's actually deployed. โ
โ โ if it can't run in prod, it doesn't exist. โ
โ โ
โ ayush@void:~$ echo $STATUS โ
โ > open to collaborations & opportunities_ โ
โ โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ


