- Spearheaded the rapid development of four generative AI proof-of-concept (PoC) applications, leveraging LLMs from Azure OpenAI to enhance automation and decision-making workflows.

- Engineered both internal and external-facing generative AI tools to optimize business operations, increasing productivity and significantly reducing manual effort across multiple teams.

- Developed robust evaluation frameworks to assess the effectiveness and reliability of generative AI applications, implementing benchmarking strategies to ensure consistent performance.

- Applied advanced prompt engineering techniques to fine-tune LLM outputs, optimizing accuracy, coherence, and consistency in financial document processing tasks such as extraction, summarization, and analysis.

- Integrated AI-driven document processing solutions using LangChain, Python, and cloud-native architectures, streamlining the extraction and analysis of financial data at scale.

- Conceptualized and implemented a scalable multi-agent system, pioneering an agentic design approach using Langchain and AWS Bedrock to solve complex, multidimensional conversational AI tasks.

- Integrated a supervisory mechanism for AI agents to adopt various personas, ensuring the optimal agent is engaged for each task segment.

- Designed system prompts for 7+ agents using advanced prompt engineering techniques, ensuring consistent agent response and actions across the system.

- Built 6 custom tools for AI agents using Python, adding new functionalities and increasing agents' performance.

- Architected the ETL pipeline for the multi-agent system using AWS, ensuring efficient data integration and management to support complex AI functionalities and system scalability.

- Constructed and managed the CI/CD pipeline using GitHub Actions, ensuring streamlined deployments and continuous integration of code changes across projects.

- Engineered an advanced Document Management System incorporating RAG with OpenAI LLMs for enhanced information retrieval and processing.

- Pioneered a Candidate Scoring System leveraging LLMs and AWS Lambda for serverless computing.

- Engineered a comprehensive system for extracting product entities and attributes from 80,000 product images.

- Utilized Azure OpenAI LLM combined with advanced computer vision techniques for precision and accurate entity extraction.

- Achieved heightened attribute detection and extraction capabilities by integrating Paddle OCR and GPT-3.5-Turbo within the LangChain framework, resulting in enhanced performance.

- Formulated consistent and clear prompts for LLMs to enhance task comprehension and efficiency.

- This innovation boosted product attribute management efficiency and customer engagement by up to 8%.

- Crafted a web-based user interface for the extraction system utilizing HTML, JavaScript, and Flask, improving accessibility and ease of use for end users.

- Partnered with MLOps and Software Engineering teams to seamlessly integrate the new extraction system into the existing infrastructure, exemplifying strong teamwork and technical coordination skills.