Abhishek Sinha

Full-Stack Developer

Tech Enthusiast who likes to self-host and build/contribute into things. Currently working on AI.

3x Microsoft Certified Professional.

AI-102, AZ-204, AZ-400

About

Full-stack web developer with almost 4 years of hands-on experience in frontend, backend, DevOps, and cloud development. Proven ability to design and deploy scalable, AI-integrated applications with GDPR complianceusing modern frameworks and cloud platforms. With a background in open source, I've evolved to build AI integrated SAAS and PAAS like Eva - The AI Assistant, NL2SQL, contributed in covid with vaccine availability bot COWIN Vaccine Availability Telegram Notifier , contributed in DuckDNS AutoIPV6forDuckDNS.

My main focus these days is integrating LLMs with custom data and refining their behavior through prompt engineering.

Experience

  1. Cloud Engineer 2

    Insight | Jan 2022 - Current

    • Built an AI-powered resume screener that reduced manual screening by 40%, using Azure AI Search and Semantic Kernel.
    • Developed a conversational AI platform using Semantic Kernel and RAG. Built with .NET backend and Angular frontend, secured by Microsoft Entra ID.
    • Developed Natural Language to SQL queries agent with Python Langchain, reducing manual API creation time by 50%.
    • Automated email parsing to extract business insights using NLP, reducing reporting time.
    • Recognized for building resume screening AI tool that saved 30+ hours/month.
    • Recognized for promoting AI tools within the SDLC, leading to a 20% increase in workflow efficiency.

  2. Subject Expert (CS)

    Chegg | Remote | College

    Helped students with their problems & questions in Computer Science Subject.

Experience

  1. Eva Assistant

    Tech Stack: Next.js, Semantic Kernel, Kernel Memory, ASP.NET, Cassandra, Redis, SignalR, Azure OpenAI, Docker

    • Developed the frontend using Next.js, Tailwind CSS, NextAuth, and TypeScript.
    • Developed the backend using ASP.NET Web API, Semantic Kernel, Kernel Memory, SignalR messaging service socket, and Apache Cassandra.
    • Migrated database management from ADO.NET to ScyllaDB for enhanced performance and scalability. Used Redis for efficient caching solutions.
    • Developed an open-source dockerized alternative to ChatGPT with Retrieval-Augmented Generation (RAG) capabilities for personalized data interaction. Implemented CI/CD pipelines for streamlined deployment.

  2. NL2SQL - Open Source Contribution

    Tech Stack: Python, Langchain, Microsoft SQL Server, Azure OpenAI

    • Using Langchain and Prompt Engineering connected a large database with LLM model GPT 3.5.
    • A user's natural language query is processed by OpenAI to convert it into SQL Syntax as per the database schema.
    • Improved response satisfaction by adding a retrieval mechanism of the database schema to the tool, which also helped to reduce the response generation time by 80% and increase stability to 98% even when used with a hospital management sample db with millions of records.

  3. Covin Vaccine Notifier

    Tech Stack: Python, Bash Scripting, API Setu APIs

    • Used API Setu publicly available government API to fetch real-time availability of vaccine shots per Indian pin code.
    • Setup Telegram Bot system to pass on data received from the script to various groups using RESTful API.
    • The solution addressed the irregular availability of vaccine slots, helping numerous people secure reservations during the peak demand period.