Performa.AI

Performa.AI is an intelligence and conversion platform for e-commerce, offering search, recommendations, product displays, cart recovery, and customer service. As a senior full stack developer and go-to engineer for technical questions, I develop and maintain APIs, catalog and order integrations, and tools embedded in stores.

Open project

Performa.AI

clients connected to the webhook service
150+
requests as a regular workload in the webhook service
Thousands/min
accesses to the API backed by replicated MongoDB
Thousands/day

Senior full stack developer and a go-to engineer for the team, responsible for building, maintaining, and improving tools and integrations.

Authenticated API, order webhook service, e-commerce integrations and platform apps, product and category synchronization, JavaScript optimizations, and AI automations.

JavaScriptTypeScriptPHPNestJSNode.jsNext.jsReactCSSMongoDBPostgreSQL

Order processing flow

Simplified view of the workflow described in this case.

  1. 1

    E-commerce platforms

  2. 2

    Webhook service

  3. 3

    Order records without duplicates

  4. 4

    Recommendation evaluation

Context

The company tools needed to access catalog data, render product showcases in online stores, and register orders from different platforms. My work covered new deliveries as well as maintenance and improvements to production systems.

Individual contribution

I built an API backed by replicated MongoDB and a service that received webhooks from platforms such as Nuvemshop, Loja Integrada, and Shopify. The service registered orders in company systems and evaluated orders associated with tool recommendations. I also implemented API integrations and apps for platform app stores.

Decision: prevent duplicate orders

Order processing needed to avoid repeated records. I implemented idempotency based on the unique order identifier to recognize orders that had already been recorded. This kept the records used to evaluate recommendations from duplicating the same order.

Decision: respect platform API limits

E-commerce platforms impose request limits on their APIs. I maintained PHP routines that queried products and categories within those limits and populated the database used by product displays. Synchronization had to account for both data retrieval and the constraints of each integration.

Decision: control API access

The API needed to serve integrations with controlled access. I implemented authentication with JWT, HMAC, and rate limiting: authentication mechanisms protect access and request validation, while call limits control consumption. The API connected to a replicated MongoDB database.

Operational scale

In the service described, more than 150 clients were connected and thousands of requests per minute were a regular workload. The API backed by replicated MongoDB handled thousands of daily accesses. These figures describe the operation during my work on these services rather than a current measurement.

Technical expertise and collaboration

I work with e-commerce platform teams to obtain information and integration guidance and support partnership agreements. I help the support team investigate issues, develop client-specific implementations, and resolve complex technical problems. I also guide early-career professionals through day-to-day technical questions and decisions.

Results and operational contribution

Catalog integrations fed product displays in stores. The webhook service recorded orders to evaluate recommendations, using unique identifiers to prevent duplicate records. JavaScript optimizations reduced tool loading times and database consumption. These deliveries supported tools designed for conversion and merchant operations; increasing sales was a product objective, with no individual commercial outcome measured in this case study.

Additional delivery: AI automations

I also worked on WhatsApp automations using AI, MCP, models, and RAG. This experience complements my work on business systems and integrations.

Company sources

Public context about the company and its platform distribution. The workloads and personal contributions described in this case study are based on my experience.