Software Engineer at Geico

Led design and rollout of microservices powering claims workflows using Node.js and TypeScript on AWS Lambda, reducing API latency by 28%. Architected a serverless event-driven pipeline with SQS, Lambda, and DynamoDB to process policy updates, delivering 3x throughput and scalable processing. Built an internal avatar-assisted negotiation prototype using OpenAI APIs and WebRTC; containerized ML inference on Kubernetes/EKS to reduce serving latency and enable horizontal scaling. Migrated legacy billing to PostgreSQL with Prisma, optimizing indexes and queries. Owned CI/CD and infrastructure-as-code with GitHub Actions and Terraform, shortening deployment time and improving rollback reliability. Instrumented services with OpenTelemetry, Prometheus, and Grafana to surface latency hotspots; strengthened data security with AWS KMS and IAM least-privilege. Mentored junior engineers and contributed a reusable React component library; designed GraphQL APIs and REST endpoints; automated ETL with

Software Engineer at resumes.fyi

Architected the full-stack resume editor and content pipeline with Next.js, TypeScript, and Node.js, cutting editor load time by 60%. Integrated LLM-driven resume suggestions using OpenAI and LangChain, increasing accepted suggestions by 48% and reducing time-to-apply. Built document parsing and normalization with Python and spaCy to handle multi-format resumes reliably. Designed PostgreSQL schema and introduced Redis caching to reduce read latency by 70% for read-heavy resume operations. Implemented real-time collaborative editing via WebSockets; containerized with Docker and deployed to AWS ECS for consistent environments. Wrote end-to-end tests with Jest and Playwright, raising test coverage to 82%. Built analytics and conversion tracking with Segment and BigQuery; created an evaluation dashboard with Faiss and Hugging Face metrics to compare prompts and embeddings. Implemented ATS export connectors via GraphQL and REST; prototyped a video-first resume walkthrough using WebRTC and F

Research Assistant at University of Houston

Led experiments on agentic LLM negotiation behaviors using PyTorch, Hugging Face Transformers, and OpenAI to simulate multi-turn negotiation strategies. Built data pipelines for large dialogue datasets with Python and Apache Beam, reducing preprocessing time by 50%. Developed evaluation metrics for negotiation success, safety, and coherence; packaged reproducible research with Docker; deployed a demo web interface for negotiation agents using FastAPI, WebSockets, and React. Mentored undergraduates on ML best practices and reproducibility.

Software Development Intern at Amazon Web Services

Implemented serverless ingestion PoC with AWS Lambda, API Gateway, and DynamoDB to support high-throughput telemetry; optimized cold starts via provisioned concurrency and modular packaging. Built authentication flows using Cognito and fine-grained IAM policies for an internal developer tool. Wrote integration tests and CI with GitHub Actions; containerized runners with Docker for consistent test environments. Collaborated across SRE and product teams to propose SDK improvements and performance tuning; presented cost and performance analysis to engineering leadership.

Research Assistant at University of Houston

Researched NLP methods for resume understanding and skill extraction using spaCy and transformer-based models. Built annotation tools and labeled datasets with React and Node.js to accelerate supervised training. Fine-tuned PyTorch models for named-entity recognition; improved F1 by 12% through data augmentation and curriculum training. Implemented embedding search with Faiss; packaged experiments with Docker and documented reproducible steps for collaborators. Co-authored internal technical reports and presented findings.

Software Engineer Intern at Buzly

Implemented social listening pipelines with Python, Twitter API, and PostgreSQL; built interactive dashboards with React and D3 to visualize sentiment and engagement trends. Developed streaming ingestion using Kafka and lightweight processors for near real-time analytics. Optimized query performance and added Redis caching to serve hot queries under 120ms. Deployed services on AWS ECS and automated CI with CircleCI; collaborated with data scientists to productionize topic classification models.