Ankit Mandal

Motivated and detail-oriented AI enthusiast with a strong foundation in machine learning, deep learning, and NLP. Proficient in Python, Flask, and Django, with hands-on experience in building intelligent applications that solve real-world problems. Passionate about image processing, computer vision, and developing scalable, user-centric solutions. Eager to contribute to innovative projects and drive meaningful impact.

AI Engineer - Prediscan Medtech - IIT madras, India (Remote)

(2024-04)

• Developed a deep learning architecture using PyTorch to detect disease severity from retinal images, implementing GridSearchCV hyperparameter tuning that boosted baseline performance by 12% and achieved an 85% diagnostic accuracy rate.
• Built advanced regression algorithms leveraging state-of-the-art vision models to calculate the Arteriovenous Ratio (AVR) and estimate cardiovascular risk (CIMT) directly from retinal scans, aiding in early disease detection for over 1,000 patients.
• Designed a high-throughput backend using FastAPI to power a multi-hospital data platform featuring a manual drag-and-drop UI, seamlessly routing large volumes of medical imagery and demographic data into a centralized inference engine.
• Engineered a secure, scalable deployment environment on Google Cloud Platform (GCP) using Docker containerization and adaptable cloud storage solutions, implementing strict data isolation protocols that selectively extract and store only bilateral retinal images, age, and gender.
• Transitioned complex research models into a production-ready Software as a Medical Device (SaMD), building a rapid, four-stage automated AI pipeline exposed via RESTful APIs to evaluate multimodal data for real-time risk assessment in active clinical workflows.

Program Analyst Trainee - Cognizant - Hyderabad

(2025-03 - 2026-01)

• Designing an intelligent test automation framework that leverages LLMs to generate Selenium test cases in Java from scenario-based inputs provided via Excel sheets
• Integrating dynamic context-sharing mechanisms to pass relevant codebase structures and logic to the LLM, enabling context-aware and accurate test case generation.
• Building a CI-compatible system to automatically execute test cases, log outputs, and utilize LLMs to validate execution logs and identify issues or false positives.
• Targeting a 60–80% reduction in manual QA effort and aiming to cut average test cycle time by over 70%, accelerating the release pipeline and improving reliability.

Machine Learning Intern - AiEnsured - Bangalore, India

(2023-06 - 2023-09)

• Conducted a comprehensive analysis of emerging software applications to implement cutting-edge administrative information systems, enhancing workflow automation.
• Developed an ethical AI model with integrated explainability for predictions, applied across object detection and NLP models, improving interpretability by 15%.
• Spearheaded the design and testing of various NLP augmentation techniques, focusing on PII protection and counterfactuals.

B.Tech - Computer Science - Shri Shankaracharya Technical Campus (2021-07 - 2025-06)