Relevant Experience

CurrentNov. 2022

Photronics Brookfield, Connecticut

Architected scalable ELT data pipeline platform for customer transaction analytics, processing purchasing data to identify optimization patterns and reduce workflow inefficiencies by 2 hours per employee weekly
Engineered automated financial statement allocation system (JSP), eliminating manual 3-week reconciliation processes and saving $60,000 annually through intelligent data processing
Developed REST API integration layer connecting legacy HR systems with Oracle EBS financial platform across international operations, reducing implementation costs by $50,000
Built intelligent purchase order automation system using Excel data extraction and validation, processing multi-line transactions and reducing manual purchasing workload by 50%

Dec. 2020Dec. 2019

Sunrisa Remote

Built data validation layer and API endpoints in Python with 10+ validation modules, using Pytest and TDD to ensure data integrity for production systems
Developed microservices architecture with gRPC and Protocol Buffers, deploying containerized applications to AWS EC2 using multi-stage Docker builds
Automated CI/CD pipelines with GitHub Actions and Makefiles, fixing critical bugs in distributed systems including timezone handling and type safety errors

Projects

Designed dual-pipeline RAG architecture separating SQL-based analytics (structured 13F holdings) from vector search (10-K narratives) using PostgreSQL, Qdrant vector database, and Claude API (Sonnet 4.5); implemented query routing logic optimizing for data type and accuracy requirements
Built scalable ETL pipeline processing SEC EDGAR filings with BeautifulSoup iXBRL parsing, dynamic text chunking (500-1000 chars), and batch embedding generation; architected cost-optimization pattern with free semantic search
Implemented metadata-filtered vector search combining 384-dim embeddings with payload indexes (CIK, section, year)
Deployed multi-cloud REST API across Supabase, Qdrant Cloud, and Streamlit with 25,210+ embeddings achieving sub-200ms query latency

Architected serverless ETL pipeline using AWS Lambda (Python 3.8), DynamoDB, S3, and CloudWatch Events with boto3 SDK to automate ingestion and processing of 1,500+ time-series records via event-driven S3 triggers
Built production REST API with Python Flask, serverless-wsgi, and AWS API Gateway performing DynamoDB partition key queries to serve JSON data—applying microservices patterns for stateless, scalable data retrieval
Automated infrastructure deployment using Serverless Framework IaC to orchestrate 5 Lambda functions, configure IAM role-based permissions, and implement rate-based scheduling for continuous data pipeline operations

20192015

Cornell University, College of Arts and Sciences Ithaca, New York

2021

22 Jul. 2021