ST

Backend · AI Engineering

Saad Tachrimant

Portfolio

Projects & case studies

A selection of backend systems, data platforms, and applied AI/ML projects. Each project highlights the problem, the system design, the technologies used, and the production engineering considerations behind the implementation.

Software systems Microservices · Integrations

Quanoni Platform

Problem: Build a scalable platform for handling consultations, payments, and notifications with high reliability and clear operational flows. System Design: Designed a microservices-based architecture using Spring Boot, Kafka, PostgreSQL, and Angular, with secure authentication and asynchronous integrations.

Implementation: Implemented OAuth2 + JWT authentication, event-driven messaging with Kafka, and third-party integrations for payments and messaging, with retries and idempotent processing where reliability mattered.

Impact: Delivered production-ready booking and messaging workflows with scalable backend services, fault-tolerant processing, and maintainable integrations.

Spring Boot Spring AI OAuth2 JWT PostgreSQL Angular Kafka WhatsApp API Payment API
AI product engineering LLM integration · Auth

Zynerator

Problem: Build an AI-assisted platform that integrates intelligent content generation into a secure, usable web product. System Design: Implemented backend services and product flows combining Spring Boot, Spring AI, Angular, authentication, and payment workflows.

Implementation: Integrated LLM APIs through Spring AI, secured user sessions with JWT, connected MySQL persistence, and implemented monetized workflows with payment provider integration.

Impact: Delivered AI-powered functionality inside a maintainable product architecture, with clear boundaries between model-powered features and core application services.

Spring Boot Spring AI JWT MySQL Angular Payment API LLM API integration
Software systems Scheduling · Platform

EngFlexy Platform

Problem: Build a reliable scheduling platform for learning workflows, instructor management, and day-to-day platform operations. System Design: Implemented backend services, integrations, and frontend components supporting scheduling and operational consistency.

Implementation: Implemented service discovery with Eureka, API integrations for payments and calendar synchronization, and frontend/admin touchpoints required for daily operations.

Impact: Delivered stable scheduling workflows and backend integrations designed for maintainability, operational continuity, and ease of extension.

Spring Boot Angular WordPress Eureka Payment API Calendar API
AI / ML · Research Edge ML · Distributed

P2P Thermal Forecasting

Problem: Build a distributed ML system that operates under real deployment constraints such as edge hardware, sensor ingestion, and limited resources. System Design: Built the full training and inference pipeline, including distributed coordination, storage, and data ingestion.

Implementation: Implemented gRPC-based coordination between nodes, time-series storage in InfluxDB, PyTorch models for forecasting, and reliable ingestion from sensor APIs.

Impact: Delivered an end-to-end ML engineering system covering data ingestion, preprocessing, training, evaluation, and inference under distributed edge constraints.

Python gRPC PyTorch InfluxDB Sensors API integration LSTM GRU
Data · AI / ML DQ · Benchmarking

THERMODSET

Problem: Create a reliable data and experimentation pipeline for thermal modeling on noisy real-world building data. System Design: Implemented preprocessing, validation, anomaly handling, visualization, and model baselines to support repeatable experimentation.

Implementation: Implemented automated data quality checks, anomaly removal, visualization, and ML baselines using XGBoost and PyTorch.

Impact: Improved data reliability and made model comparisons reproducible through structured preprocessing and benchmark-oriented experimentation.

Python PyTorch XGBoost LSTM Data visualization Quality checks Anomaly removal