Powering the future with cloud-native infrastructure and AI-assisted learning. Master Generative AI, Kernel Security, VLSI design, and Edge Systems with live tools and real production workflows.Explore Data Science Research →
AI Systems Research Track
Production AI Systems
Engineered for 2026
A technical deep-dive into the stack: from Transformer internals to Agentic workflows and Inference optimization.
The Production AI Stack
Architecture
- Transformers
- MoE
- SSMs
- Multi-modal
Fine-tuning
- LoRA / QLoRA
- FSDP
- DeepSpeed
- Unsloth
Inference
- vLLM
- TensorRT-LLM
- SGLang
- Speculative Decoding
Retrieval
- Graph RAG
- Vector DBs
- HyDE
- Re-ranking
Agents
- LangGraph
- AutoGen
- Toolformer
- Swarm
Evaluation
- LLM-as-Judge
- Arena-Hard
- LiveBench
Alignment
- DPO / RLHF
- Constitutional AI
- Red Teaming
Deployment
- MLflow
- LangSmith
- BentoML
- Modal
Engineering Learning Path
- 1
Foundations
- Transformer Architecture
- PyTorch Internals
- Tokenization & Attention
- 2
Optimization
- FlashAttention-3
- Quantization (GGUF/AWQ)
- Mixed Precision Training
- 3
RAG & Memory
- Vector Embeddings
- Semantic Search
- Knowledge Graphs
- 4
Agentic Systems
- Reasoning Loops (ReAct)
- Multi-agent Orchestration
- Function Calling