Interactive Labs

Matmul, softmax, attention, KV cache, training, CPU mechanics, sampling.

Original deep-dives into transformer internals (tokenization, attention, QKV, softmax).

Temperature, top-k/p, KV cache memory, speculative decoding, attention masks.

Multi-head attention, positional encoding, FFN, MoE routing, RMSNorm.

Embedding similarity, gradient descent, k-NN, decision boundaries, RAG pipeline.

FP16/INT8/INT4, SmoothQuant outliers, GGUF format, perplexity vs bit-width.

Message flow, tool registry, capability negotiation, server lifecycle, OAuth.

Raft election, gossip spread, vector clocks, quorum reads/writes, partition tolerance.

Token ring, consistency levels, compaction strategies, tombstones, hinted handoff.

Rate limiter, circuit breaker, load balancer, cache patterns, sharding.

B-tree, MVCC, replication lag, connection pooling, pgvector ANN.

Kafka partitions, consumer lag, exactly-once, stream-table duality, backpressure.

Producer-consumer, lock contention, work stealing, race conditions, virtual threads.

SLO burn rate, percentile latency, trace sampling, alert fatigue, golden signals.

Token bucket, ring buffer, consistent hashing, TCP congestion, ABR ladder.

TCP/TLS handshakes, BBR vs CUBIC, anycast routing, DNS resolution.

OAuth+PKCE, JWT decoder, mTLS, SQL injection demo, password hashing.

WebSocket frames, HTTP/2 multiplexing, HOL blocking, SSE, reconnect backoff.

Jitter buffer, VAD, LUFS targets, echo cancellation, sample rate conversion.

ABR decision, codec comparison, HLS timeline, bitrate ladders, WebRTC pipeline.