Shreeya Bekkam

Machine Learning Engineer (New Grad, May 2026) with 2+ years' experience building and deploying ML, NLP and Deep Learning based systems in production settings. Developed end-to-end pipelines across data processing, modeling, and API integration on GCP. Focused on delivering practical, scalable ML solutions by working closely with engineering and product teams.

Machine Learning Engineer - Vosyn - USA

(2025-09 - 2026-03)

• Architected and deployed a production-grade speech AI pipeline using WhisperX (ASR + forced alignment) and NVIDIA NeMo (speaker diarization) as GPU-accelerated, containerized microservices on Vertex AI; enabled automated speaker-labeled transcription at scale and exposed services via a FastAPI-based API gateway on Cloud Run with GCS-backed I/O and a structured JSON interface for modular, independent scaling.
• Designed and implemented a CI/CD pipeline using GitHub Actions with multi-stage quality checks (Flake8, pytest, Docker → Google Artifact Registry), streamlining deployment across Vertex AI and Cloud Run and reducing release cycle time by 30%.

Data & Applied ML Engineer - SaferTek IT Solutions - Hyderabad

(2023-01 - 2024-07)

• Built end-to-end data pipelines in Python (Pandas + SQL) consolidating sales, inventory, procurement, and product data across UWAREONE modules; resolved schema inconsistencies, cleaned legacy records, and aggregated weekly/monthly datasets used for downstream demand forecasting and planning.
• Developed a forecasting workflow using lag features, rolling averages, and moving average trend models on historical sales data; integrated predictions into backend services using Flask APIs and Tableau dashboards, reducing manual planning effort by 35% and lowering stock mismatch incidents by 15%.

Research Intern - Genoparadigm, KMIT - Hyderabad

(2022-01 - 2022-12)

• Designed and fine-tuned a VGG-16 CNN-based computer vision pipeline for malaria parasite detection on annotated blood smear images; implemented YOLOv5 for object detection, improving diagnostic accuracy through systematic data preprocessing and model optimization.
• Prototyped GCN-based (Graph Convolutional Network) multi-stage classification pipelines for parasite localization — applied exposure to graph theory and symbolic graph structures for real-world biomedical classification.

Master of Engineering - Artificial Intelligence - University of Cincinnati (2024-08 - 2026-05)

Bachelor of Technology - Computer Science and Engineering (AI & ML) - Jawaharlal Nehru Technological University (2020-10 - 2024-05)