Harshad kale

Built an AI-powered sales automation system using Python (FastAPI) to generate personalized email and LinkedIn outreach at scale. Built WebSocket-based real-time preview systems for AI-generated email and LinkedIn messaging, allowing users to view and refine content instantly. Designed event-driven AWS workflows using SQS, Lambda, Step Functions, EventBridge and ECS. Integrated MongoDB for scalable storage of user profiles, campaign data, message templates, and AI-generated content. Designed and optimized AI prompts for personalized email and LinkedIn sales outreach. Actively investigated and resolved customer-reported issues, collaborating across teams to improve system reliability, content quality, and user experience.

Developed a Secure Access Portal using Django, Django REST Framework, and JWT-based authentication, enabling secure, role-based access control and fine-grained authorization for sensitive user data. Led full-cycle backend development by customizing authentication backends, enforcing session security policies, and integrating middleware-level activity monitoring to harden the application against common vulnerabilities. Designed and maintained SKALA, a full-stack Task/Project Management System built with Django and Django REST Framework (DRF), delivering robust backend APIs backed by a MySQL relational database. Developed an interactive frontend using React, implementing a dynamic drag-and-drop task board, API-driven views, and client-side state management for improved usability and responsiveness. Built modular, scalable features for User Authentication, Project Management, and Employee Management, implementing role-based access and ensuring cross-device responsiveness and performance.

Built a web application Question Answering Engine using Haystack framework. 25 pretrained models were tested on different aspects (score, time, memory, free/ paid, accurate answers). After testing models, GPT3 (Text-davinci-003) model was selected to be used for our project. Developed a Better Word Cloud and Knowledge Graph system using NER, pyvis, and REBEL model, storing outputs in MongoDB. Utilized React for frontend development.