Most viewed

Day 1: Tokens, Context Windows, and Why Models Forget Mid-Task3 views · 1 like
Python Type Hints Complete Guide: Write Self-Documenting, Bug-Free Code3 views
Python Decorators Explained: From Simple Wrappers to Production Patterns3 views
System Design Interview Cheat Sheet: The Framework That Gets You Hired at FAANG3 views
Day 9: Eval Harness Day One2 views · 1 like

Newly added

CheatCoders

API Gateway WebSockets for Multi-Turn Coding Agents: Connection State Without Sticky Myths

IDE coding agents need multi-turn WebSockets — store ConnectionId in DynamoDB, route $connect/$disconnect/default, fan-out via the callback API, watch Lambda concurrency, and resume with sessionId after reconnect. Skip sticky ALB myths; pair with response streaming for one-shot HTTP.

AI AWS Debugging

CodeBuild Spec Sandboxes for AI Coding Agents: Ephemeral Builds Without ECS Sprawl

When an AI coding agent must run tests or lint on a PR branch, a CodeBuild project is a cheaper build-centric sandbox than ECS Fargate sprawl. Use buildspec, privileged false, optional VPC, S3 artifacts, and an IAM role scoped to the repo bucket.

AI AWS System Design

KMS Decrypt Grants for Agent Tools: Least Privilege Without Env Keys

Agents need secrets for tool APIs — do not put plaintext in Lambda env. Use Secrets Manager + KMS grants and encryption context (tenantId, toolName), constrained grant tokens, and CloudTrail audits. Never kms:Decrypt on *.

AI AWS System Design

EventBridge Scheduler: Overnight Agent Batches Without Cron Drift

Nightly RAG reindex, eval, and batch Converse jobs drift when cron rules fight DST and retries. EventBridge Scheduler gives one-time vs rate schedules, flexible time windows, Asia/Kolkata timezones, DLQs, and idempotency tokens so missed runs do not double-apply.

AI AWS Debugging

Lambda Recursive Loop Protection: Stop Agent Self-Invokes From Burning Quotas

Coding agents that InvokeFunction or fan-out Lambdas can recurse forever and burn concurrency quotas. AWS recursive loop detection stops same-account Lambda-to-Lambda cycles — wire circuit breakers, maxDepth on tool schemas, DLQs, and CloudWatch alarms before an agent calls itself via alias.

AI Machine Learning

Day 1: Tokens, Context Windows, and Why Models Forget Mid-Task

Tokenization, context windows, and recency bias silently break chatbots — measure token budgets and attention to the middle before you pick a model or RAG design.

AI Machine Learning System Design

Day 2: Transformer Attention as a Systems Diagram

Treat transformer attention, KV cache, and batching as a production data path — latency, memory, and throughput come from systems choices, not paper diagrams.