Applied AI/ML Engineer at Tabner Inc. working on KonfigAI (No-Code GenAI Platform), building production-grade AI applications using RAG pipelines, LLMs, and scalable backend systems.
I design and develop end-to-end AI systems, including document intelligence platforms, data-driven analytics applications, and AI-powered automation tools. My work focuses on integrating machine learning models, building efficient data pipelines, and deploying scalable backend architectures.
I specialize in solving real-world problems using Generative AI, natural language interfaces, and intelligent data systems. Currently, I’m focused on AI agents, LLM orchestration, and building high-impact production systems.
My professional journey and key projects I've worked on
• Multi-Tenant RAG: Architected a multi-tenant RAG platform (DocPortal) using FastAPI, Weaviate, and hybrid retrieval, enabling 50+ enterprise users to query 1,000+ documents.
• Search Optimization: Improved retrieval accuracy from 58% to 85% using HNSW Hybrid Search, metadata filtering, and retrieval evaluation pipelines.
• SQL Agent: Developed an LLM-powered Text-to-SQL assistant (Pioneer Nexus) for PostgreSQL analytics, reducing analyst query creation time by 40%.
• HR Automation: Orchestrated Azure-based cloud data pipelines for an AI resume optimization system, reducing candidate screening time by 50%.
• Production Systems: Managed Kubernetes-based production deployments on GCP with CI/CD automation, maintaining 99.9% system availability.
• Open Source: Maintained an active open-source footprint of multi-agent templates, reducing team-wide project setup time by 30%.
• Data Engineering: Performed feature engineering and exploratory analysis on 5GB+ retail datasets using Python and SQL.
• Dashboarding: Designed Power BI dashboards reducing executive report generation time by 30% and improving decision visibility.
During my internship at DataValley, I worked on web development projects using Python, Flask, and Django. I developed dynamic web applications, integrating HTML, CSS, and JavaScript to create responsive and user-friendly interfaces.
View Certificate
• Medical Vision: Trained a YOLOv8 computer vision model achieving 94% mAP in identifying skeletal fractures.
• Medical Chatbot: Rolled out an NLP chatbot delivering 98% accurate drug side-effect data.
As part of my college project, I led a team of 6 members in developing a Streamlit web application that translates documents into various languages. The project allowed me to showcase my leadership skills and work collaboratively with a diverse team.
View ProjectOver the course of a month, I collaborated with a friend to create visualizations for various data structures such as stack, queue, linked list, array, circular queue, and priority queue.
View ProjectHere are some of my notable projects showcasing my skills and expertise
A production-grade multi-agent research pipeline that structures search intents, ranks and evaluates source credibility, resolves source conflicts, and maps dynamic network graphs.
An end-to-end AI automation workflow that handles content research, multi-agent post generation, and automated publishing with AI-generated visuals.
A data-driven pharma analytics platform delivering descriptive, predictive, and prescriptive insights. Includes ML-based sales forecasting, demand-expiry matching, and an AI chatbot that converts natural language queries into dynamic SQL for real-time data retrieval.
A RAG-based document intelligence system that enables users to manage and query multiple PDFs within a single project using metadata filtering and relevance scoring for accurate information retrieval.
A Django web application for automated code error detection with 50+ quality rules and an AI explanation engine that reduced developer debugging time by 40%.
Interactive visualizations for various data structures like stack, queue, linked list, and more.
Computer vision project implementing object detection algorithms for real-world applications.
Streamlit application that translates documents into various languages with a user-friendly interface.
A RAG-powered legal assistant using Mistral 7B and LangChain to query 120+ pages of legal documentation with custom chunking, achieving 92% semantic search precision.
Here are the key technical skills I've acquired through my academic and professional journey
Insights, tutorials, and thoughts on AI, technology, and development
An in-depth look at building an asynchronous multi-agent system that structures user search intent, scores credibility, resolves source conflicts, and maps interactive knowledge graphs.
Explore how I built an end-to-end AI-powered LinkedIn automation workflow using n8n, Gemini, and custom Cloudflare AI APIs.
A technical deep dive into DocPortal AI — covering hybrid search, multi-tenant vector storage, async ingestion pipelines, and multi-LLM orchestration with Groq and Gemini.
How I built a pharma analytics platform combining ML-based sales forecasting, demand-expiry intelligence, and a natural-language-to-SQL AI chatbot for real-time data retrieval.
Certifications and achievements that validate my skills and knowledge
Feel free to reach out for collaborations or just to say hello!
tusharahul82@gmail.com
Rustumbada, Narsapur, Andhra Pradesh, India
tusharahul.netlify.app
6304747097
Welcome to my channel! We're time-traveling to the future, one daily vlog at a time. The destination? Me, in five years. The journey? Probably filled with questionable choices.
The subscribe button is down there somewhere. Feel free to ignore it.