12 ideas and counting

Idea Library

Real AI engineering builds with full build environments. Pick one, build it, ship it.

12 ideas

Builder

RAG-001 · RAG & Knowledge

RAG orchestrator with self-healing retrieval logic

Production retrieval systems fail silently. Build an orchestrator that detects poor retrieval quality and reroutes queries using fallback strategies — before the user sees a bad answer.

2 Weeks
View brief
Builder

AGT-002 · Agents

Stateful AI agent framework with reflective memory

Most agent frameworks can't remember what they did two steps ago. Build a stateful agent runtime with episodic memory, reflection loops, and rollback capability.

2 Weeks
View brief
Advanced

LLM-003 · LLM Infra

LLM serving layer with dynamic KV cache

Serve LLMs efficiently at scale. Implement dynamic KV cache management, request batching, and latency-aware routing across model replicas.

1 Month
View brief
Foundational

DAT-004 · Data & Features

Real-time prediction drift detection system

ML models degrade in production without warning. Build a streaming drift detector that compares live feature distributions against training baselines and alerts on deviation.

1 Week
View brief
Builder

SAF-005 · Safety

LLM safety and alignment evaluation framework

You can't deploy what you can't evaluate. Build a modular safety harness that runs adversarial probes, jailbreak tests, and alignment checks across any LLM endpoint.

2 Weeks
View brief
Builder

VEC-006 · Vector & Search

Vector database indexer with HNSW

ANN search is at the core of every RAG system. Implement HNSW from scratch, expose a REST API, and benchmark it against production vector databases.

2 Weeks
View brief
Builder

RAG-007 · RAG & Knowledge

Multi-hop question answering pipeline

Single-retrieval RAG breaks on complex questions. Build a pipeline that decomposes multi-hop queries, retrieves across sub-questions, and synthesizes a grounded answer.

2 Weeks
View brief
Foundational

AGT-008 · Agents

Tool-use orchestrator with structured output validation

LLM tool-calling is brittle without validation. Build an orchestrator that wraps tool calls with schema enforcement, retry logic, and structured error recovery.

1 Week
View brief
Builder

LLM-009 · LLM Infra

Prompt cache layer with semantic deduplication

Most LLM apps re-compute answers to semantically identical prompts. Build a caching layer that stores completions by semantic similarity, not exact hash.

2 Weeks
View brief
Advanced

DAT-010 · Data & Features

Feature store for ML pipelines

Training/serving skew kills production models. Build a lightweight feature store with point-in-time correctness, online serving, and offline batch access.

1 Month
View brief
Foundational

SAF-011 · Safety

PII detection and redaction pipeline

User data leaks through LLM prompts constantly. Build a preprocessing pipeline that detects and redacts PII before it reaches any model endpoint.

1 Week
View brief
Builder

VEC-012 · Vector & Search

Hybrid search engine with re-ranking

Dense retrieval misses keyword-critical queries. Build a hybrid search system that blends BM25 and dense embeddings, then re-ranks with a cross-encoder.

2 Weeks
View brief

More ideas added every month. Paid members get instant access.