Jose Miguel Alves

Senior Software Engineer

15+ years of experience building AI-powered systems and large-scale data infrastructure. My work sits at the intersection of machine learning, distributed systems, and cloud engineering — from multi-agent LLM applications to Spark and Kubernetes workloads on AWS. Some of it has ended up published or in a patent application. I care about shipping things that actually work in production.

Skills

AI / LLM
LangGraph LangChain FastAPI Multi-Agent Systems RAG LangFuse
ML & Data
TensorFlow Keras Scikit-Learn Pandas Spark MLlib AWS SageMaker OR-Tools
Big Data
Apache Spark Kafka Hive Presto AWS EMR DynamoDB PostgreSQL ClickHouse
Cloud & Infra
AWS Terraform Kubernetes Docker CI/CD EKS Lambda Step Functions
Languages
Python TypeScript Scala SQL Node.js Java
Architecture
Microservices Event-Driven Distributed Systems RESTful APIs MLOps

Experience

2022 – Present

Varicent

Senior Software Engineer — Data & ML Backend · Toronto

Built a production multi-agent AI cost analysis assistant (LangGraph, FastAPI) that identified ~40% in monthly AWS savings across several accounts. Led a major Spark rearchitecture using a DAG-based execution framework and Kubernetes orchestration, cutting pipeline execution times by over 50%. Also designed Monte Carlo simulation and linear optimization tools using OR-Tools, AWS Lambda, Step Functions, and EKS, and replaced ad-hoc infra scripts with Terraform reducing deployment times by 30%+.

2019 – 2021

Paladin AI — acquired, now PACE InstructIQ (TXT Group)

Founding Software Engineer — Data & ML Backend · Montreal

Built core parts of an AI-powered pilot competency platform: event-driven serverless architecture on AWS processing real-time simulator data, end-to-end ML services (LSTM, Random Forests on SageMaker and Spark MLlib) for flight maneuver segmentation and performance grading, and a multi-region data lake (DynamoDB, S3, Kinesis, Glue, EMR) handling terabytes of training data. Terraform across all ML pipelines and CI/CD workflows. Mentored backend engineers on MLOps and AWS patterns.

2017 – 2019

Western University

M.Sc. Researcher & Teaching Assistant · London, Canada

Designed and built ML4IoT, a microservices and Docker-based framework for orchestrating parallel ML pipelines on IoT data — published in IEEE Access. Built Spark applications in Python, Scala, and Java Spring Boot to stream IoT data through Kafka into HDFS. Taught as TA in Databases (×2) and Software Construction (×2).

2010 – 2017

Ícaro Technologies — IBM Business Partner

Software Engineer — Data Analytics & ITSM · Campinas, Brazil

Built custom analytics products for major Brazilian telecoms using ML models (Decision Trees, Linear Regression, Association Rules) to predict network outages and surface fault patterns affecting 60M+ customers. Designed a near-real-time data warehouse on Netezza (Snowflake schema, DataStage pipelines) that reduced KPI processing time by 98.95%. Earlier, deployed and customized BMC and IBM ITSM solutions and built integration architectures for Network Management products.

Open source

AWS CLI Configure

VS Code extension

A VS Code–compatible extension for Cursor, VSCodium, and any Open VSX–based editor. Open your ~/.aws/credentials and ~/.aws/config files directly from the editor, switch the active AWS profile from the status bar, and set any named profile as [default] — no terminal required.

Education

2017 – 2019

Western University

M.E.Sc. in Software Engineering · London, Canada

Thesis. Research focus on ML frameworks for IoT data pipelines, published in IEEE Access.

2005 – 2010

University of São Paulo — USP

B.Sc. in Information Systems · São Carlos, Brazil

Publications & Patents

Apr 2021

USPTO Patent

Automatic inferential pilot competency analysis based on detecting performance norms in flight simulation data

US20220335850A1 →

Oct 2019

IEEE Access Journal

ML4IoT: A framework to orchestrate machine learning workflows on Internet of Things data

ieeexplore.ieee.org/document/8876834 →

Contact

Interested in collaborating or have a question? Reach out via email or LinkedIn.