Shruti Sawant

ProfessionalTechnical Support Specialist

The environment. A globally deployed SaaS assessment platform used by enterprise recruiting teams: browsers, IDE environments, coding-assessment engines, REST APIs, databases, and CRM/ticketing (HubSpot, Jira Service Management). During live events, one failure is visible to thousands of candidates at once.

My role. I owned complex cases from the inbound queue and live channels. I reproduced failures, gathered evidence, found which layer was failing, and escalated to Engineering, Product and QA with reproduction steps, logs, environment details and severity, while keeping the customer conversation until resolution.

Events supported in real time

• Google The Big Code: 95K+ participants
• Cepheid Hiring Challenge: 87K+
• Shadowfax Data Warriors: 46K+
• CodeSecure 2025: 23K+
• Meesho Hackathon: 20K+

Beyond the queue. Partnered with Product to reproduce reported security concerns and validate AI-assisted proctoring across browsers; recommended workflow and browser-compatibility improvements; wrote knowledge-base articles and runbooks that cut repeat escalations.

Open Source Scholar & Contributor Dan Kohn Scholar

In plain terms. Kubernetes runs applications in containers. Kubescape, a CNCF security project, checks those clusters for risk. I built a port and service-discovery scanner for it that finds which services (Cassandra, MySQL, PostgreSQL, Redis) are listening, and improved network scanning efficiency 10×.

Before: network scanning
After: port and service-discovery scanner, 10× efficiency

Awarded the Dan Kohn Linux Foundation Scholarship (2.2% worldwide acceptance) to attend KubeCon + CloudNativeCon North America, Oct 2022, where I worked with Kubernetes, Envoy service mesh, Prometheus/Grafana and CI/CD.

LeadershipTeam Leader · 6 members

Top 5 nationally, of 299,352+ students.

Problem. Users with disabilities need communication tools that work in real time (AICTE, problem statement PS1464).

What we built. A real-time video-conferencing platform on WebRTC, Socket.io, Node.js and React, with JWT/bcrypt authentication, plus an AI chatbot and Python voice assistant. I led the team through delivery, Jul – Dec 2023.

ResearchAccepted

Simply put. You describe a scene in words; the model generates a video of it. NeuroVidX uses diffusion models guided by an expert transformer.

Technical detail. Python, PyTorch, TensorFlow, Transformers; trained with deep neural networks; deployed with Docker and scaled on AWS.

Accepted at the International Conference on Advancements in Computing Technologies (AI Track). Forthcoming in Advances in Intelligent Systems Research (Springer Nature / Atlantis Press). Publication details. Link

Machine-learning models for sales trends reaching 90% forecast accuracy; Power BI dashboards for KPIs and churn; dynamic-pricing scenarios for profit margins; Tableau reporting automation that cut reporting time by 40%. Python, Pandas, NumPy.