Building distributed systems and observability tools. Delivered a full-stack DRAM visualizer (131k+ datapoints/frame), fault-tolerant Azure/Kubernetes pipelines (99% deployment success), and gRPC telemetry in Rust/C++. Automated deployments (-30% lead time, $50k/yr savings) and improved reliability with Prometheus/Grafana (1.5× fewer outages, 50% faster image pulls).

• Built full-stack interactive DRAM visualizer using React, Express, and D3.js to render 131k+ dynamic data points/frame for hardware chip diagnostics; architected a performant frontend with lazy loading and quadtree spatial indexing algorithm, and oversaw end-to-end system integration for real-time analysis.

• Architected & developed a multi-region web app using Python flask + JS to visualize deployment performance metrics from our custom datacenter; engineered data flow from Azure Cold Storage to containerized services deployed via Kubernetes, ensuring 99% deployment success with a fault-tolerant design and seamless failover support.

• Instrumented gRPC with C++ and Rust to enable fine-grained telemetry, metrics, & better observability for apps and introduced certificate management, reducing security risks by 30% & executing seamless gRPC workloads.

• Automated deployment process by creating a new pipeline to custom cloud using Azure Service Principal, Ansible & Python, producing unified k8s deployment images, decreasing lead time for change by 30%

• Integrated Prometheus-based monitoring & alerting in K8s cluster via IaC, reducing outages by 1.5×, and

hooked Slack-API to forward alerts using Python. Optimized container image pulls by configuring Azure Container Registry cache rules, reducing DockerHub pulls by 90% and lowering pull costs; with efficient caching, cut average pull times by 50%.

• Technologies: Rust, C++, gRPC, Python, Flask, React, Javascript, Node, Docker, Kubernetes, Azure (Table, Blob, Data Lake, Service Principal), Ansible, Jinja, Grafana, Prometheus, SSL, TCP, Git.

Software Development Engineer at Amazon — Built full-stack web tools (Java, React/Redux, AWS) scaling to 3,000+ sites/400+ markets; created hiring simulator (~$3M savings). Led Operational Excellence with AWS CDK dashboards + CloudWatch automation, cutting SEV turnaround from 10+ hrs → 10 mins.

• Developed full-stack web tools using Java (Spring Boot, Smithy) and React/Redux to support site feasibility analysis, competitor intelligence, and real-time updates across 3,000+ sites and 400+ markets; empowered retail expansion and planning teams with actionable insights at scale.

• Created modular features—messaging systems, report generators, and executive approval workflows—backed by NoSQL (MongoDB, DynamoDB) and SQL (Amazon RDS) and deployed via AWS ECS, EC2, Elastic- Search, CloudFront, SNS.

• Built an optimized hiring simulator enabling strategic experimentation with location and workforce levers, driving an estimated $3M annual cost savings.

• Led Operational Excellence by building an AWS CDK dashboard visualizing 12 health metrics on EC2-powered resources; authored automated CloudWatch alarms that triggered SEV tickets via Slack-API—reducing issue turnaround from 10+hrs to 10 mins.

• Added accessibility using React with dark mode features to the hiring simulator web app according to WAG standards and end-to-end playwright tests

• Technologies: Java, Spring, Smithy, React, Redux, JavaScript, TypeScript, Jest, Vite, AWS (API Gateway, RestAPIs, EC2, SNS, SQS, DynamoDB, MongoDB, RDS).

Developed NoC simulators, optimized LLMs with compression techniques, and built AWS Batch workflows—while creating an open-source 3D ML Ops visualizer for large-scale models. Hands-on with Python, PyTorch, React, Docker, AWS, and advanced performance optimization.

• Built a NoC simulator for analysis, modeling traffic on a 2D grid and applying path traversal algorithms (A*, Dijkstra, weighted, round robin) to evaluate latency and power/cost efficiency.

• Optimized a BERT(LLM)-based NLP system for real-time inference by applying model compression techniques (pruning, early-exit, distillation, quantization), enabling production-ready AI features with minimal accuracy loss.

• Built AWS Batch workflows: job definitions, queues, compute environments, ECR/ECS integration, and an SNS

queue to trigger worker scripts upon DynamoDB updates.

• Built an Open Source 3D ML Ops visualizer (Python Flask, ThreeJS, D3.js) using quadtree spatial indexing

to render and interact with large-scale models (Eg.MNIST, ResNet, BERT) data in real time.

• Technologies: Python, PyTorch, JavaScript, TypeScript, React, Docker, Bazel, AWS, ThreeJS, D3JS.