Skip to content
Mani Chokkara

Selected work

Systems, written down.

Each of these is a full case study: the problem, the architecture, the engineering challenge, the decision, the alternative I rejected, and what happens when it fails.

Flagship platform

AURYNT Data Factory

Enterprise synthetic data infrastructure

A cloud-native, multi-tenant platform that generates, validates, certifies and delivers realistic synthetic datasets — built so teams can develop, test and demo against data that behaves like production without touching production.

2025 — ongoing
  1. 01Schema intakeTenant schema, constraints, business rules
  2. 02GenerationStatistical + behavioural generators
  3. 03ValidationDistribution, referential, rule checks
  4. 04CertificationSigned quality report per dataset
  5. 05DeliveryBatch, incremental and streaming outputs
Read the case study

Streaming system design

Real-Time Fraud Detection

High-throughput real-time decisioning

A streaming architecture that scores transactions in flight — ingest, feature computation, model inference and decisioning on the hot path, with everything durably landed for analysis and retraining.

2025
  1. 01TransactionsProducers emit keyed events
  2. 02KafkaPartitioned by account for ordering
  3. 03Stream processingWindowed aggregates, enrichment
  4. 04FeaturesLow-latency online feature reads
  5. 05InferenceModel scores the enriched event
Read the case study