← Engineering Intelligence Hub Himanshu Jain · Director of Software Engineering LinkedIn
GraphLens
Architecture Intelligence
Navigation
GraphLens Home Architecture Map AI Insights 6 Impact Analysis Platform Architecture Platform Insights Impact Simulator AI Copilot
Repositories
📊 Repository Explorer 10
Case Studies
📚 Architecture Case Studies 5
Platform Architecture Intelligence
Live Service Interaction Map
Real-time visualization of service dependencies, data flows, and cross-service communication patterns.
GAPI Lazlo ML Toolkit dynamic _pricing incentive -service PCC JTier message -service subscription -programs PCC UI
Pricing Flow
Incentive Flow
ML Pipeline
Subscription Flow
💰
Pricing Flow
GAPI / Lazlo  — consumer request
dynamic_pricing  — GET /bulk/currentPrice/v2
Pricing Control Center JTier  — ILS schedule
VIS / Federated Inventory  — mBus events
ML Toolkit  — min-price floor → Pricing Service API
📨
Campaign Messaging Flow
Tetris CMS  — campaign create
message-service  — audience assign + schedule
incentive-service  — /event/notify qualification
CAS Arbitration  — inbox throttle + rate
Rocketman / Regla  — email + push delivery
🧠
ML Prediction Pipeline
📊Hive/Janus
Events
🔄Airflow
DAGs
🧮Bayesian
STAN
📈Elasticity
Table
💲Pricing
Engine
Bayesian Elasticity Model
STAN hierarchical model estimates deal-level price elasticity → median_bayesian_elasticity table
Min Price Base Pipeline
MDS + EDW + Teradata → Hive/TEZ → Pricing Service API min-price floor
Location Embedding Model
Janus events → BigQuery → GCS Parquet → location embedding training
🔗
Cross-Service Dependencies
incentive-service
AMSMDSPricingCassandra
message-service
incentive-svcRTARocketman
dynamic_pricing
VISMySQLmBusML Toolkit
subscription-programs
incentive-svcKillBillOrders
PCC JTier
Pricing SvcVISHiveQuartz
ML Toolkit
AirflowSageMakerVertex AIBigQuery
AI Engineering Insights
Automated Architecture Analysis
GraphLens continuously analyzes graph topology to surface architectural risks, bottlenecks, and optimization opportunities.
⚠️
HIGH RISK · Dependency Fanout
Pricing engine has critical dependency fanout
dynamic_pricing has 13,521 dependency edges — 47% of total edge count. Single-service blast radius affects TPIS, VIS, PCC JTier, and message-service simultaneously. Recommend circuit breaker pattern and fallback pricing cache.
dynamic_pricingincentive-servicePCC JTier+4 more
🎯
CRITICAL · Runtime Dependency
Incentive service is a critical checkout runtime dependency
incentive-service has 22,346 nodes and is called synchronously in the checkout flow. subscription-programs-app's #1 community (78 nodes) is entirely dedicated to calling incentive-service APIs. Any degradation propagates directly to order conversion.
checkout flowsubscription-programsmessage-service
🔁
MEDIUM · Retry Amplification
message-service may amplify retry storms under load
6 background jobs across batch/app fleets run independently (no Redis lock). Campaign Logger + ImportAudienceActor share Cassandra write paths with no distributed coordination. Concurrent fan-out to Rocketman can cause retry cascades during peak campaign sends.
message-serviceCassandraRocketman
🧠
INFO · ML Coupling
ML toolkit directly influences dynamic pricing calculations
Bayesian elasticity model outputs flow into pricing optimization. Min-price DAG feeds directly to Pricing Service API. Any drift in Hive upstream data quality or Airflow DAG failures causes silent price floor miscalculation.
machine-learning-toolkitdynamic_pricingAirflow DAGs
📐
MEDIUM · Layer Violation
REST resource layer bypasses service layer in subscription-programs
SelectResource calls Membership domain model directly (INFERRED, 0.85 confidence). Violates service-layer pattern — any Membership model change requires direct REST resource update. 197-degree hub node confirms widespread coupling.
subscription-programs-appSelectResourceMembership model
HEALTHY · Layer Separation
dynamic_pricing shows clean service layer boundaries
Graph shows clear community separation: HTTP layer (Request, degree 142), DAO layer (ProgramPriceStateDAO), service layer (CurrentPriceService), and mBus publish layer. Exception filter chain is the only legitimate cross-cutting concern bridging all layers.
dynamic_pricingCurrentPriceServicemBus
Cross-Repository Impact Analysis
Change Blast Radius Estimator
Understand which services are affected when a repository changes. Used for release planning, incident triage, and architectural reviews.
💲
dynamic_pricing
Tier-1
Change impact on →
🎁
incentive-service
CRITICAL
🌏
pricing-control-center-ui
HIGH
🔄
subscription-programs-app
HIGH
🧠
machine-learning-toolkit
MEDIUM
🎁
incentive-service
Tier-1
Change impact on →
🔄
subscription-programs-app
CRITICAL
📨
message-service
HIGH
🌙
pricing-control-center-jtier
MEDIUM
💲
dynamic_pricing (API)
MEDIUM
🧠
machine-learning-toolkit
Data
Change impact on →
💲
dynamic_pricing (min-price)
CRITICAL
🌙
pricing-control-center-jtier
HIGH
📨
message-service (deal feed)
MEDIUM
🎁
incentive-service (sellout)
MEDIUM