Alex Vall

Async RAG (AI) application to digitalise any insurance documents following regulatory rules & using different LLMs depending the use case.

• End To End python AI implementation & solution.
• Async digitalisation using Celery & Redis as message broker.
• Haystack, Langchain, Weaviate, Flowise, FastAPI, Celery.

PaaS (Platform As a Service with Multi-tenant strategy) to simplify the relation between Distributors, Brokers, Manufacturers & Final Customers.

• Lead the team to implement the foundations (control plane) of the platform.
• Design the multi-tenancy architecture of the platform taking care:
  • Business OKR's to adopt the best solution in terms of performance & costs. The PaaS is a business based on economy of scale.
  • Document RFC's to adopt the best solution in terms of performance, scalability, resilience and security.
  • Data isolation in tenant level using Postgres-RLS (Row Level Security)
  • Tenant & Identity context propagation using an extra header in the request
  • New authorisation approach decoupled from authentication using custom new Permission API & Open Policy Agent
  • SDK library to ensure consistent multi-tenancy adoption in all the domains.
  • Migration plan to adopt multi-tenancy pool model solution
  • Database migrations
  • Broker migration
  • SDK library adoption
  • AWS infra definition to support multi-tenancy
  • EKS (4 clusters) as container manager using Karpenter. Ingress per domain as Load Balancer.
  • AWS API Gateway as Public gateway using lambda authorizers
  • Spring Cloud Gateway as domain gateways
  • RDS Postgres (as database) applying CQRS pattern (synchronized replicas to read/write)
  • Redis as distributed cache
• Implementation of the foundations library & design of the platform architecture:
  • API services using Hexagonal to ensure decoupled modules.
  • Foundations library using Starter patterns & covering reactive (WebFlux).
  • Database Schemas & RLS (definition & modifications) under Liquibase.
  • Apache Debezium as CDC to inject data into the Data Lake using Kafka connectors and Kafka Streams.
  • JOLT to transform JSON's.

Technologies used:

• Java 17
• Spring Framework (Boot, Cloud)
• Python (AWS lambda authorizers)
• AWS RDS Postgres (RLS - Row Level Security)
• Redis (distributed cache)
• OPA (Open Policy Agent)
• AWS Lambda / Step Functions
• AWS API Gateway
• AWS EKS
• JOLT
• python 3
• Haystack
• Langchain
• Weaviate
• Flowise

Lead the Claims team to accomplish these goals:

• New Claims API using ontology in the OpenApi definition of the damage (coverage extension) syntax definition.
• Automatic fraud detection using AI.
• Automatic car claim evaluation using AI.
• Deprecate legacy payout system & design automatic integration with finance.
• Google AI Dialogflow to facilitate the user’s input of claims.

Lead the Heroku Connect (sync Salesforce data with SQL database) deprecation project.
Design & implementation of a cheaper custom solution to reduce the high costs of Heroku Connect.

• Bidirectional sync between Salesforce and RDS Postgres.
• Kafka as a broker to work in async way.
• Apache Debezium with kafka connector (as CDC) to synchronize from Postgres to Salesforce.

Technologies used:

• Java 17
• Spring Framework (Boot, Cloud Stream)
• AWS Lambdas
• AWS Step Functions
• Apache Kafka (KStreams, KSql)
• Apache Debezium
• Google Dialogflow Chatbot (AI)

API first adoption as a methodology to work.

• Automatic code generation for DTO’s & API controllers.
• “Feature branch” as strategy to reach the fastest way to work with multiple developers.

Design a common Hexagonal Architecture for all the Microservices implementations.

• Archetype for Maven lifecycle tool.

Microservice architecture (design and implementation) for ‘Backend For Frontend (BFF)’ to provide services to Mobile / Web Apps.

• GraphQL API to connect Mobile and Web Apps with BFF.
• gRPC (protobuf) to connect BFF against Core Microservices.

Microservice architecture (design and implementation) for backend core to provide business rules related with the new Insurance Company called ONE.

• Step Functions and Camunda BPM for orchestration and Business Processes. Using BPMN notation. Claim coverage validation, Fraud detection. Calculate the costs of any claim.
• gRPC (protobuf) for synchronous communication between Microservices.
• Apache Kafka as a broker to work in async way.

Technologies used:

• Java 17
• Spring Framework (Boot, Cloud Stream)
• GraphQL (Apollo)
• RDS Postgres
• Apache Kafka (KStreams, KSql)
• Camunda BPM
• gRPC (protobuf)

IOS Mobile App for Switzerland market to facilitate the switch process between Insurance Companies.

• Viper architecture as a design pattern (View, Interactor, Presenter & Router).

Monolith architecture (design and implementation) for Backend For Frontend.

• Design taking care the OKR’s & low budget.
• RabbitMq for async operations.
• Enterprise Java (Spring Extension) Common library for all microservices. Simplifying dependencies.
• Decoupled microservices architecture into same Github repository.

Frontend for Backoffice (offers).

• AngularJS as a framework.

Technologies used:

• Swift
• Cocoapods
• AngularJS
• Java 11
• Spring Framework (Boot, Cloud Stream)
• Spring Integration
• Heroku Connect (Sync between Salesforce & Postgres)

B2B IOS Mobile App (ipad) for Monsanto (South Africa market).

• Viper architecture as a design pattern (View, Interactor, Presenter & Router).
• Possibility to work offline. Background synchronization.
• Salesforce integration using a Java middleware (Backend For Frontend).

BCA (British Car Auctions).
Shoppers use the app to confirm bets placed by the moderator.

• IOS Mobile App (ipad) to automate car auctions. Online App using custom gestures.
• Viper architecture as a design pattern (View, Interactor, Presenter & Router).

BCP (Central Bank of Peru).
IOS Mobile App (ipad, iphone) that automates all daily tasks of any end user.

• Viper architecture as a design pattern (View, Interactor, Presenter & Router).

Technologies used:

• Swift
• ObjectiveC
• Cocoapods
• SQLite
• Java 11
• Spring Framework (Boot)
• Spring Integration
• Salesforce

B2B IOS Mobile App (ipad) for Colomer (Revlon - cosmetics).
Revlon’s sales team uses the app to sell in hairdressing salons.

• Viper architecture as a design pattern (View, Interactor, Presenter & Router).
• Possibility to work offline. Background synchronization.
• SAP integration using a Java middleware (Backend For Frontend).

Backend Application (Technotree) - Helsinki (Finland).
Messaging Server for Telecommunications Operator (Swisscom, Sonera).

• Voice mailbox (iphone), Video mailbox, SMS, Streaming, Alarms, Notifications
• Developed using JAVA open source.

Technologies used:

• Swift
• ObjectiveC
• Cocoapods
• SQLite
• Java 8
• Spring Framework
• JAX-WS-RS (cxf)
• Hibernate
• JBOSS
• ESB (under JBOSS) as process orchestrator.