Maintainer of Burn, an open-source deep learning framework written in Rust. Contribute across the full stack including ops, backends, ONNX import, WASM, `no_std`, and performance optimization.

Built and maintain burn-onnx, an ONNX model importer for Burn with broad model coverage, codegen, and GPU/CPU correctness. Designed and implemented burn-store, Burn's model weights import/export system with zero-copy loading, quantized tensor storage, and mixed-precision support.

Applying Burn to real-world speech recognition and ASR pipelines, focusing on fast, portable ML inference that runs across servers, WASM, and embedded devices.

As part of the Alexa Help Domain (Alexa virtual support) group, developed, designed and maintained the group’s core services to support . Actively participated in architectural discussions and reviews. Hired, developed and mentored new developers. Prioritized and planned projects with key stakeholders.

• Designed and developed the new workflow recommendation subsystem of the AHD platform.

• Rewrote the business workflow management system to a newer and faster React based UI.

• Migrated Mayday’s communication component to the new AWS Connect backend.

• Reduced operational costs by 100K yearly by optimizing resource allocations for existing service.

• Created analysis tool to reduce case troubleshooting time to several minutes from hours.

• Reduced the connection latency and error rates to the Mayday’s service.

Worked as a full-stack developer in an agile environment, participating in the software development lifecycle, from prototyping and designing to releasing. Analyzed requirements from a development perspective and decomposed them into functional and implementation tasks with high-level effort estimates.

• Developed system for automatically extracting data from unstructured text from inbound emails.

• Designed framework to support MLS data transformers in a declarative manner.

• Developed new IDX connectors for downloading Multiple Listing Service (MLS) properties

Led the development team and mentored developers of all levels on different aspects of technology and development. Worked with product management to understand business requirements and create technical specifications, as well as determining realistic project schedules and resource estimates. Designed, developed, and tested architectural components and frameworks. Peer-reviewed code and provided feedback on design, performance, style, and maintenance issues.

• Integrated the Exemption Certificate Management (Certs) product with Sales Tax Decision services.

• Overhauled the search and report capabilities of Certs to meet growing performance demands.

• Led a major rewrite project of converting Certs to Python and migrating the entire stack to AWS.

• Designed and developed the next generation of Returns Engine to unify the existing, disparate, legacy Sales Tax Return processing system and reduce response time from minutes to milliseconds.

Optimized and scaled the company’s main product for a large number of concurrent users. Ensured the application’s high availability and stability. Audited the codebase for potential security vulnerabilities. Automated manual processes to improve developer productivity and reduce errors.

• Developed and architected Exemption Certificate Management web app from the ground up. It grew to become the company’s main product, used by big clients, such as Intuit and Apple.

• Developed and launched scalable web service for generating PDF exemption certificates on the fly.

• Exposed the product’s functionality via REST APIs for client’s CRM integration.