Learning path

Technologies

Databases, cache, queues, search, AI

  1. Key technologies roadmap One problem, one failure mode, one 'don't use this for' — per tool. Postgres, DynamoDB, S3, Elasticsearch, queues, streams, distributed locks, Redis, and CDNs, run through a single film-crew analogy so the roles stay distinct.
  2. PostgreSQL Interview-default SQL: social-platform example, B-tree/GIN/GiST/covering/partial indexes, WAL write path, replication, transactions and row locks, OCC, partitioning, sharding — plus hands-on Docker lab and when to leave Postgres.
  3. MongoDB Document model, BSON, full mongosh + Node driver hands-on, indexes, embed vs reference, aggregation, replica sets, sharding, read/write concern, transactions, and change streams — DynamoDB-depth for interviews.
  4. Apache Cassandra Wide-column data model, partition keys, consistent hashing, replication, QUORUM, LSM writes, gossip — plus query-driven modeling with Discord and Ticketmaster examples.
  5. Amazon DynamoDB Fully managed key-value store: partition and sort keys, GSIs/LSIs, query vs scan, per-request consistency, RCU/WCU math, DAX, and Streams — plus when to pick it in an interview.
  6. Time-series databases How TSDBs achieve 10–100× throughput on metrics workloads: append-only storage, LSM trees, delta/XOR compression, time partitions, Bloom filters, rollups, tag indexes — and when Postgres is still the better choice.
  7. Caching strategies Where to put a cache (CDN, Redis, client, in-process), the four read/write patterns, eviction and TTL, then stampede, consistency, and hot keys — plus how to introduce caching on the whiteboard without jumping straight to Redis.
  8. Redis Why Redis earns a deep dive: in-memory data structures, cluster hash slots, cache and locks, leaderboards, rate limits, streams vs Pub/Sub — and when Kafka or Postgres should win instead.
  9. Message queues When to go async, delivery guarantees, and the failure modes that surface first — poison messages, consumer lag, and exactly-once as a marketing term for effectively-once.
  10. Apache Kafka World Cup stats as motivation: producers, partitions, consumer groups — then a comparative guide to Redis queues/pub-sub/streams, SQS, Kinesis, RabbitMQ, ActiveMQ vs Kafka, plus cost/performance levers.
  11. Apache Spark Batch analytics deep dive: Driver/Executors, DAG stages and shuffle, deploy modes, executor memory, spark-submit and CLI commands, PySpark/SQL hands-on — plus when Spark beats Flink.
  12. Spark Structured Streaming Micro-batch streaming on Spark: architecture, readStream/writeStream, triggers, watermarks, output modes, foreachBatch, Kafka exactly-once, ops commands — and when to pick Flink instead.
  13. Apache Flink When stream processing beats batch: dataflow graphs, keyed state, watermarks, windows, JobManager/TaskManagers, checkpoint barriers for exactly-once — Redis dashboards, fraud CEP, and interview when-to.
  14. Elasticsearch Documents, indices, mappings, and the REST API — then Lucene segments, inverted indexes, shards, and query planning. When Elasticsearch earns a box on your diagram and when Postgres full-text is enough.
  15. Vector databases Embeddings, similarity metrics, exact vs approximate nearest neighbors, HNSW/IVF/LSH indexing, filtered and hybrid search, pgvector vs Pinecone, hot/cold indexes, and when vector search earns a box on your diagram.
  16. Prompt engineering Production prompt engineering for system design interviews: anatomy, techniques, structured outputs, RAG prompting, tool loops, safety, evaluation, architecture patterns, worked examples, and cost levers.
  17. RAG architectures Principal-level RAG deep dive: ingest, chunking, hybrid retrieval, reranking, grounded generation, multi-tenant ACL, index lag SLOs, embedding migrations, evaluation harnesses, end-to-end worked examples, and interview scripts for knowledge Q&A systems.
  18. Agentic application architectures Principal-level agentic architectures: tool loops, plan-and-execute, routers, multi-agent supervisors, durable workflows, memory layers, tool safety, sagas, blast-radius isolation, evaluation, worked support-agent design, and interview scripts.
  19. Agentic patterns Catalog of agentic design patterns with worked examples: ReAct, plan-and-execute, reflection, router, orchestrator–worker, map–reduce, supervisor, RAG-as-tool, Self-RAG, HITL, memory hygiene, pattern economics, and composition playbooks for production systems.
  20. Agentic frameworks Survey of agentic frameworks with principal-level selection criteria: LangGraph, LlamaIndex, CrewAI, provider SDKs (OpenAI, Claude, Google ADK, Microsoft), Pydantic AI, Mastra, lock-in trade-offs, migration playbooks, and interview decision trees.
  21. Deep agents — LangGraph & LangChain Interview-style Q&A: how Deep Agents (harness), LangChain (framework), and LangGraph (runtime) fit together — plus nine types of agentic applications you can build, with pick-lists, cost levers, and Bollywood/politics analogies.
  22. Context management with LangGraph Context engineering for agents: trim, summarize, offload, retrieve, and isolate — and how LangGraph's state, checkpointers, and store make these patterns easy in production (including the UI-vs-LLM history split).
  23. Hosting agentic apps on AWS Production reference for running LangGraph / Deep Agents on AWS: Fargate vs EKS vs Lambda, SQS async workers, Aurora checkpointers, S3 offload, Bedrock, VPC/IAM sandboxes, HITL, observability, cost levers, and interview scripts.
  24. API Gateway Hotel front desk for microservices: request routing, auth, rate limits, and the six-step request flow — plus when a gateway is overkill for a monolith.
  25. Kubernetes A clear roadmap: control plane, how replication heals, Deployments/Services/Ingress, HPA, a short kubectl lab, and when K8s belongs on your diagram.
  26. Apache ZooKeeper A clear roadmap: why coordination is hard, ZNodes + watches + ensemble, the four patterns you build, ZAB in brief, and when to pick ZooKeeper vs etcd/Consul/KRaft.

Lattice