Project atlas
240 catalog entries · showing 51–100. 1 published so far.
Data Lineage Catalog
Trace datasets and fields across ingestion, dbt, APIs, and models; show ownership, freshness, and impact analysis.
Time-Series Telemetry Store
Design partitions, retention, downsampling, compression, and queries for high-cardinality service metrics.
Vector Index Benchmark Lab
Compare pgvector and Qdrant HNSW/IVF configurations on recall, latency, memory, ingestion, and filtering.
Cache Correctness Playground
Demonstrate cache-aside, write-through, stampede protection, stale-while-revalidate, and invalidation failure modes.
Backup and Point-in-Time Recovery Drill
Automate backups, restore to a target timestamp, verify integrity, and document measured RPO/RTO.
Tenant Isolation Testbed
Compare schema-per-tenant, database-per-tenant, and row-level security with migration and noisy-neighbor tests.
Tamper-Evident Audit Ledger
Create append-only events with hash chaining, signatures, retention, verification, and privacy-aware redaction.
Data Retention and Deletion Engine
Enforce policy-based TTLs, legal holds, export, deletion propagation, and verifiable tombstones.
Search Relevance Workbench
Tune BM25 analyzers, stemming, synonyms, field boosts, and offline relevance judgments for a real catalog.
PostgreSQL CDC Replication Pipeline
Capture changes with Debezium, evolve schemas, deduplicate, replay, and validate source-to-sink consistency.
Product Clickstream Platform
Define an event taxonomy, manage sessions and identities, process late events, and expose trustworthy funnel and retention tables.
Lakehouse Table Format Lab
Compare Parquet plus Iceberg/Delta concepts for partitioning, compaction, snapshots, schema evolution, and time travel.
Idempotent Airflow Backfill System
Build parameterized DAGs with retries, backfills, data intervals, atomic outputs, and safe partial reruns.
dbt Marketplace Warehouse
Model facts, dimensions, SCD Type 2 history, incremental models, tests, documentation, and metric definitions.
Data Quality Gatekeeper
Enforce schema, freshness, volume, distribution, referential integrity, and business rules before downstream promotion.
Real-Time Anomaly Stream
Detect spikes, drops, missing partitions, and distribution shifts with event-time windows and explainable alerts.
Search Indexing Pipeline
Transform source records into Elasticsearch/OpenSearch documents, handle updates and deletes, and support zero-downtime reindexing.
Recommendation Feature Pipeline
Generate offline and online features from behavioral events, prevent leakage, and verify point-in-time correctness.
Data Contract Registry
Let producers declare schemas, semantics, owners, SLAs, and compatibility; show downstream impact before changes merge.
PII Discovery and Governance Pipeline
Classify sensitive fields, propagate tags through lineage, enforce masking, and audit policy decisions on synthetic data.
Pipeline Observability Console
Track freshness, completeness, lineage, run duration, retries, cost, and incident history across batch and streaming jobs.
Feature Store Reference Stack
Materialize consistent offline/online features, version definitions, monitor skew, and serve low-latency lookups.
Self-Service Analytics API
Expose governed metrics with caching, row-level security, query limits, semantic definitions, and reproducible exports.
Spring Boot Service Generator
Generate Clean Architecture modules, tests, OpenAPI, Docker assets, migrations, and CI from a validated specification.
Go API Code Generator
Generate typed handlers and clients from OpenAPI, preserve manual code safely, and prove determinism with golden tests.
OpenAPI Breaking-Change Detector
Classify contract changes, comment on pull requests, and provide migration guidance with fixture-based accuracy tests.
Database Migration Linter
Flag unsafe locks, table rewrites, missing rollback plans, and non-concurrent indexes for PostgreSQL migrations.
Secret and Credential Scanner
Combine entropy, patterns, allowlists, and commit-history scanning; report precision and recall on a labeled corpus.
SBOM and License Inspector
Produce CycloneDX output, detect risky licenses and vulnerable packages, and enforce configurable merge policies.
Structured Logging Kit
Provide Python, Java, and Go libraries with correlation IDs, redaction, sampling, JSON schemas, and cross-language examples.
Reproducible Benchmark Harness
Control warmup, concurrency, datasets, and confidence intervals; export comparable HTML reports.
Feature-Flag SDK Trio
Implement matching Python, Java, and TypeScript SDKs with deterministic bucketing, caching, offline fallback, and contract tests.
GitHub Review Automation App
Enforce repository-specific quality checks, summarize actionable failures, and maintain a complete audit trail.
Local Code Search Engine
Index repositories with lexical and semantic search, incremental updates, symbol filters, and evaluation queries.
Cross-Platform Task Runner
Build dependency-aware tasks, parallel execution, caching, cancellation, and cycle detection.
Stateful API Mock Server
Generate mocks from OpenAPI, model workflows and faults, record traffic safely, and support deterministic replay.
Docs-as-Code Portal
Generate versioned API and architecture docs, validate links and examples, and preview every pull request.
Astronomical Image Classifier
Build careful splits, augmentation, transfer learning, error analysis, uncertainty, and an experiment ledger tied to astro-imaging experience.
Document Layout and OCR Pipeline
Detect regions, extract text and tables, preserve coordinates, evaluate by document type, and support human correction.
Speech Transcription and Diarization Service
Segment audio, identify speakers, transcribe, timestamp, redact test PII, and report word/diarization error rates.
Visual Defect Detection System
Compare classification, localization, and anomaly approaches under scarce labels, augmentation, and domain shift.
Time-Series Transformer Forecasting Lab
Compare transformer, boosting, and statistical baselines with rolling validation and latency/cost analysis.
Multimodal Catalog Matcher
Align image and text embeddings, retrieve duplicates/substitutes, evaluate hard negatives, and expose an explainable review tool.
Tokenizer Engineering Suite
Implement BPE, train on domain corpora, inspect fertility and edge cases, and compare vocabulary, speed, and compression.
Small GPT from Scratch
Implement attention, masking, position encoding, training, sampling, checkpoints, evaluation, and an interactive architecture explainer.
Domain LoRA Fine-Tuning Study
Curate and version data, train LoRA/QLoRA adapters, compare against prompting/RAG baselines, and publish a model card.
Quantization Quality Lab
Compare int8/int4 and GGUF/AWQ/GPTQ-style workflows on memory, throughput, latency, and task-quality regression.
vLLM Production Endpoint
Serve an open model with continuous batching, streaming, quotas, autoscaling signals, OpenAI-compatible APIs, and load results.
Offline llama.cpp Assistant
Run a quantized model locally with privacy controls, hardware-aware configuration, and benchmarks across context sizes.
Speculative Decoding Experiment
Implement or integrate draft/target decoding, measure acceptance and speedup, and document when overhead loses.