Feb 2024 – Aug 2025 · 1 yr 6 mos · Gurugram, Haryana, India · Hybrid

• Developed end-to-end semantic search using Hybrid search (combination keyword-BM25 and neural search-KNN based search) via Opensearch on AWS.
• Reranking (RRF ranking) & Dynamic Clustering of relevant questions for Gartner’s Magic quadrant for vendors in a specific market.
• Designed solution for faster GenAI deployment cycle using Langchain and postgres. Deployed in the form of an endpoint as well as an internal package.
• Implemented a Retrieval-Augmented Generation (R.A.G.) based application for Gartner Associates, including the migration of documents to a Vector Database (ChromaDB) for RAG consumption.
• Conducted integration with multi-agent LLM software and prompt engineering for diverse document use cases, such as rewriting documents for a new role,summarization, key findings extraction, and quality assessment.
• Developed similarity scoring algorithms (Perplexity scoring) to compare Generative AI content against various non-AI inputs, facilitating accurate ranking.
• Deployed and created APIs on AWS lambda and internal packages that can be utilized across various teams and verticals, thereby accelerating AI adoption and leveraging Generative AI capabilities organization-wide.
• Built an end-to-end application for various VPs to visualize data science POC’s use cases via a Streamlit interface, hosted on AWS SageMaker.

May 2022 – Dec 2023 · 1 yr 7 mos

• ● Implementing an end-to-end deployment solution for dynamic pricing(Hotels & Flights) and demand
• forecasting(Hotels) has shown significant promise. Results indicate a 4.9% boost in revenue, and further optimization
• has the potential to drive a substantial 7% increase in Gross Book Value (GBV), resulting in even higher revenue.
• ● Enhancing and optimizing the Search Ranking Page (SRP) algorithm for both homes and hotels to boost Clickthrough
• Rates (CTR) and Conversion Rates (CVR) significantly. Utilizing the Falcon WSGI framework for API.
• ● Developing a User Clustering pipelines for various verticals, including Flight, Hotel, and Location, while adhering to
• SOLID principles. This approach ensures code reusability across diverse projects and promotes maintainability,
• scalability, and efficiency in development.
• ● Contributed to Research Paper for Context-based Cross-selling Recommendation System for Online Travel Agencies.
• Optimized MongoDB upsertion and data generation using Dask, while planning ML pipelines in GCP with Kubeflow
• and Kubernetes.
• ● Creating an end-to-end pipeline for an automatic flight email parser is designed to significantly reduce operational
• costs by replacing manual processes for filtering and delivering flight updates to customers.
• ● Integrating monitoring API/APM services via DataDog.

Jun 2021 – Apr 2022 · 10 mos

• Designing communication pipeline for API listeners using Apache Kafka Queue(pub-sub).
• Worked on models like Face detection, gaze estimation, facial recognition & activeness. Worked on end-to-end AI powered online proctoring system design and analysis software.
• Hands-on experience in building automated Question & Answering (Q&A) systems, leveraging Transformer-based models and Natural Language Processing (NLP) pipelines. Additionally, I have expertise in developing text summarization solutions. During the entire Software Development Lifecycle (SDLC), I have successfully optimized AWS server storage consumption, significantly reducing the storage load required for model creation and deployment.

Jun 2020 – Jun 2020 · 0 mo