Automating FindingFive’s deployment workflow by integrating Poetry, Ansible, CI/CD pipelines to deliver secure, reliable, and optimized updates.

Setting up automation for configuring temporary, feature-specific test environments with secure, ephemeral URLs

Integrated New Relic APM to give the company visibility into transaction performance and error traces, reducing MTTR by 30% and strengthening security by remediating 4 High and 12 Medium CVEs.

Implemented a custom Slack-based logging system that integrates with New Relic and Falcon APIs, with real-time error reporting, decreasing customer-reported incidents by 70% through proactive monitoring and automatic ticket creation.

Students have to click multiple times to observe a menu that is serving for today. Students want to track calories but there is no option to do so.

mydininghall.com is a student-focused application, where we are continously working and building based on feedbacks of users to provide a smooth dining lookup experience.

Secured $20K investment to scale platform into multi-tenant architecture for deployment at different universities (SUNY Oswego, Vanderbilt University), providing school-agnostic data pipelines and white-label infrastructure.

Designed full-stack platform serving 70+ daily users using Django, Next.js, PostgreSQL, and Amazon Web Services (EC2, RDS, S3) with 99\% uptime.

Implemented full observability stack with Grafana Loki for centralized logging, Prometheus for metrics collection, and Grafana dashboards for visualization.

Built distributed automation infrastructure using Celery + Redis task queue for web scraping 1000+ daily menu items, integrated with Nginx and Cloudflare for production deployment.

Observability-Aware Routing for AI Inference

• Investigating whether observability-aware routing reduces tail latency in AI inference systems

• Building testbed with multiple inference backends instrumented with Kubernetes, Prometheus and Grafana to compare routing strategies

• Building SUNY Oswego Interactive Maps (Current) (Technology used - React, Tailwind, MapLibre) to showcase SUNY Oswego locations/features in one place. (e.g. parking locations and services, building info)

• Provide geospatial analytics through web application (Technology used - React, Tailwind, GMaps, Turf.js) for assisting in the policy decision making of SUNY Oswego commuter distance

Working under Department of Enrollment Management

Developed a logistic regression model using Bloomberg market indicators, back-tested on 19 years of data, outperforming the baseline strategy by 30%.

Crafted an LLM evaluation app to benchmark the performance of 3+ models simultaneously, leveraging GroqCloud for ultra-low-latency inference