Experience

• 

  Solarillion foundation

  Apr 2020 - Oct 2022

  Research assistant

  • Developed a 2-stage pipeline to predict the flight delay using XGBoost Classifier and Regressor with R2 score of 0.95 and F1 score of 0.95 for class 0 and 0.78 for class 1 prediction.• Developed BP-Net, an end-to-end DL solution that uses PPG waveform to estimate SBP, MAP and DBP through ABP waveform, satisfying BHS and AAMI criteria with MAE of 5.16 mmHg and 2.89 mmHg for SBP and DBP• Deployed BP-Net on an RPi 4 with 4.25 ms inference time to translate the PPG to ABP waveform.

• 

  Lti - larsen & toubro infotech

  Aug 2021 - Apr 2022

  Senior software engineer

  • 5 months training on Full stack development and Java 11 Oracle Certification• Training under Citi Bank with Metadata correction and quality analysis, since January 2022

• 

  Optisol business

  Apr 2022 - Jul 2023

  Machine learning engineer

  • Developed a forklift safety device using transfer learning on Efficient-Net to detect Person and Industrial objects.• Optimized the pipeline in accordance with OpenVino to run 4 camera inference simultaneously with 30 FPS and accuracy of 79% on Ubuntu OS and Intel NUC and alert based on different zones of danger• Deployed a Nuclei Segmentation pipeline for WSI Images to count the total number of positive and negative contours to identify the cancer cells in the given image.• Built a data model and analyzed the flow of data using different visualization tools like Segment, Braze, etc., and gave personalized recommendations for users across the app. Show less

• 

  Arizona state university

  Jan 2024 - now

  Graduate student assistant

  Implemented movement generation pipeline for IARPA-HAYSTAC project to generate instructions for agents for 4 AOIs, based on learned rules using PyReason.Utilized A* algorithm to generate movement instructions adhering to event, time, and duration constraints, satisfying the Objective Satisfaction Score of 1.Identified anomalous locations with Isolation Forest, providing feedback to refine the generated path.