Keerthanyaa Nirmalkumar

As a Junior Generative AI Developer at CGI, I work at the intersection of software engineering and artificial intelligence, with a focus on the financial services domain. My responsibilities include: • Designing and refining prompts for large language models (LLMs) • Developing and managing data pipelines • Deploying AI solutions that enhance financial reporting and client communication tools • Aligning AI capabilities with business goals in a collaborative, client-facing environment

* Collaborating with AI engineers, developers, and data scientists to build scalable, intelligent applications. * Developing and deploying AI-powered systems for data ingestion, model execution, inference pipelines, and visualization. * Building cloud-native services with containerized microservices and serverless architectures. * Implementing full-stack features across backend APIs (Python, Flask/FastAPI) and modern frontends (React, TypeScript). * Working with large-scale datasets in DynamoDB, MongoDB, and relational databases. * Exploring advancements in cloud computing, AI agents, ChatGPT integration, and DevOps tooling.

As a Substance Use Peer Educator (SUPE), I was part of a certified peer education program focused on promoting harm reduction and fostering a safer, healthier campus environment. I contributed to the development of outreach materials and supported educational events and classroom presentations aimed at increasing substance use awareness among students.

Built and deployed ML models and NLP pipelines for transcript summarization, OCR-based structured data extraction, and NL-to-SQL conversion on real-world, unstructured enterprise data Applied statistical and feature engineering techniques to optimize NL2SQL prompt strategies — reducing token usage and improving SQL generation accuracy across multiple benchmark schemas Designed multi-agent AI workflows using LangChain and LangGraph for automated orchestration, integrated across AWS (Lambda, Bedrock, EC2) and Azure (OpenAI, Document Intelligence) Operated in agile sprint cycles, translating client business requirements into structured data science solutions and presenting findings directly to stakeholders