Ankit Chauhan

- Developed and deployed a full-stack knowledge graph application with a 36-endpoint FastAPI backend, Cytoscape.js visualizations, and a responsive React/TypeScript UI for exploring complex community ecosystems. - Fine-tuned BERT to classify asset vs.\ deficit language, expanding a 270-example seed set into a 1,350-example labeled corpus; deployed as a containerized Flask service (F1 = 0.68, p95 latency = 3.4s).

- Implemented data pipelines for an NSF-funded project to ingest and preprocess educator survey data, curate LLM training datasets, and fine-tune models (QLoRA, DPO). Final model achieved a 14% improvement in pedagogical alignment and an 8% reduction in hallucinations over baseline. - Designed culturally responsive annotation guidelines and QA workflows for K-12 instructional content dataset of 7,200 items, achieving an inter-annotator agreement of Cohen's kappa 0.88. - Supported over 30 students in the Fall 2024 H518 Deep Learning course through tutorials, labs, office hours, and lesson planning on topics: GANS, CNNS, LSTMs, Transformers, and Reinforcement Learning.

Treasurer of the Graduate and Professional Student Government (GPSG) at Indiana University Indianapolis - Elected as the treasurer of Graduate Professional Student Government (GPSG) to represent student interests and manage a budget of $85,000. - Revamped the sponsorship and funding request process increasing student organization’s fund utilization to 85% (up from 55% in prior years). - Served as co-chair to IU Funding Board, oversaw allocations of $230,000 in grants to organizations. - Recipient of Graduate Student Honors award for Academic Excellence while maintaining leadership role.

- Engineered a data ingestion pipeline using PySpark to load data into 100+ Hive tables, reducing data processing time by 40%. - Built an XGBoost based predictive forecasting model that optimizes SKU selection for enterprise licensing needs, delivering 18% annual cost savings on underutilized license spend for 43,000+ users. - Developed Cloud Functions to process and load over 10 GB of daily raw JSON data from GCS bucket into BigQuery and Cloud SQL. - Led migration for legacy applications to Azure cloud environment during an enterprise split ensuring service continuity for over 23,000 users.

- Designed and maintained CI/CD pipelines in Azure DevOps to automate testing and deployment of internal business and operations applications to Azure App Services, cutting deployment time by ~30%. - Authored 20+ knowledge-base documents on routine administration tasks and led regular knowledge sharing sessions on topics storage lifecycle management, retention policies, and Azure CLI best practices across EMEA, APAC, and NA teams.