Aug 2022 – Present · 3 yrs 7 mos · Bengaluru, Karnataka, India

• Chatbot and Excel Assistant-
• Developed a scalable RAG pipeline using Parquet-based storage, cutting data load time by 60%.
• Integrated a Pinecone vector Database with Cohere Reranker, Metadata filtering and BGE embeddings to build semantic indexes, achieving 85% retrieval accuracy and saving 50K USD annually in triage costs.
• Fine-tuned Llama 3.1 8B using QLoRA on 150K+ customer support tickets, achieving 94% classification accuracy and reducing response time by 70%
• Built a smart Excel plugin enabling users to Rephrase, Summarize, and Highlight data (cell/row/sheet), seamlessly integrated with Excel UI using LLMs.
• Achieved a 45% reduction in OpenAI token usage per query and 30% faster response time by minimizing data scope through intelligent filtering
• Incident Automation
• Collaborated with cross-functional teams to develop an NLP based Automated Incident Resolution System, processing over 1M incidents across 61 global service lines saving 100K USD annually.
• Implemented a One-vs-Rest classifier to categorize newly clustered incidents, optimizing DBSCAN epsilon value to 0.7, achieving 82% accuracy.- Automated workflows using Apache Airflow, reducing manual intervention by 40%.
• Cloud Infrastructure Services
• Automated ETL workflows and optimized MSSQL databases for 200+ batch processes, reducing incident response time.
• Improved system reliability by integrating Control-M for batch job scheduling and ensuring 90% server uptime.

Mar 2022 – May 2022 · 2 mos · Bengaluru, Karnataka, India