Software Engineer with 5+ years of experience building production systems, automation workflows, data pipelines, AI agents, and full-stack products for SaaS, B2B, fintech, manufacturing, and internal operations.
I like to develop reliable software that makes work faster, decisions smarter, and systems easier to operate.
- Led the redesign of an existing manufacturing quotation feature using React, TypeScript, Django REST Framework, and PostgreSQL. Defined requirements with Operations, designed the interface, APIs, and database schema, and implemented most of the changes. Assigned tasks and reviewed the team’s work.
- Built a FastAPI CAD-processing service with Azure Service Bus and Container Apps, supporting up to 35 concurrent jobs per worker replica, with retries and failed-message handling.
- Independently built the Node.js LogoDetection worker on AWS, combining OCR, LLM classification and configurable PDF redaction. Built a LangGraph PPAP workflow used by operations, reducing estimated report preparation from a working day to five minutes or less.
- Developed a manufacturing assistant with LangChain, LangGraph, Azure OpenAI, Django APIs and PostgreSQL/pgvector for technical retrieval and permission-checked quotation actions. Used Azure AI Foundry to deploy models and configure endpoints.
- Replaced deterministic cost estimation with ML pricing using PyTorch geometry features, XGBoost regression, scikit-learn evaluation and FastAPI inference.
- Built Databricks ETL pipelines with real manufacturing data, PySpark, Spark SQL, bronze/silver/gold Delta tables, scheduled jobs, data-quality checks and SQL dashboards.
- Built GitHub Actions CI/CD for automated tests, linting, Docker builds and deployment through Azure Container Registry to Container Apps. Managed production Kubernetes workloads on Azure with Terraform and Grafana, plus scoped cloud identities, permissions and secrets.
- Deployed the LogoDetection worker through Amazon ECR/ECS with CloudWatch monitoring. Secured S3 file access with permissions and expiring signed links, and removed unused Azure resources to reduce unnecessary spending.
- Improved PostgreSQL models through staged migrations and backfills, and resolved N+1 queries that reduced Orders page loading from about five seconds to one.
Started in support and moved into software engineering responsibilities within approximately three months.
- Built financial-data integrations using Python, PowerShell, SQL Server, FIX transactions, XML mappings and daily CSV reports over FTP/SFTP, from provider requirements and historical loads to production reconciliation.
- Wrote stored procedures, views and functions to transform and validate 10,000+ transaction records. Resolved a doubled account total by tracing an unexpected summary row in a provider file and correcting the filtering rules.
- Investigated client issues through application logs, SQL records and incoming files. Worked directly with clients and providers in English, documented findings and mentored junior engineers.
- Built a Flask/ORM tool to simplify repetitive database operations, and used Bloomberg Terminal and its Desktop API to investigate security identifiers and price data.
- Built Laravel/Livewire ERP modules for fleet, tires, maintenance, invoicing and employee-performance workflows using PHP, JavaScript and MySQL.
- Developed tournament management, user accounts, role-based permissions and reusable interface components.
- Gathered requirements from operational staff and built Python/SQL ETL, Excel reporting and Power BI dashboards.
I use Codex and Claude Code to plan features, define expected behavior and tests, implement changes and review the result, followed by my own code inspection and validation. I also use Postman collections, environments, authentication flows and scripted assertions for API testing, with collections run manually.
- Clean, maintainable systems with a clear source of truth
- Production reliability, observability, and simple operations
- AI tools that solve real workflow problems instead of adding noise
- Fast iteration without hiding errors or complexity
- Products that feel useful, polished, and grounded in user needs
