Abhisek Dey Email & Phone Number

Rochester, Ny, Us

Rochester, New York, United States

Continuing my work in Chemical Data Extraction and Entity Linking

South San Francisco, California, United States

Extraction and Parsing of Molecule Diagrams from chemical patents and journals

Fairport, New York, United States

Created an Asynchronous Parallelized Training Data Pipeline for creating aligned depth map projections along with synchronized and registered RGB and Thermal Images. Further reduced processing times by 600%Solely delivered a proof-of-concept project on the substantial performance improvement on a supervised real-time model using the processed depth maps above on the thermal domain (Thermal Supervised Monocular Depth Estimation).Developed a distributed (multi-node, multi-GPU) training and evaluation pipeline for Depth Completion on sparse LiDAR data based on a multi-domain recurrent model. Demonstrated marked improvement of fully-filled depth maps over traditional interpolation techniques. (To be used in production)Collaborated with Machine Learning Lifecycle stakeholders to make all my projects are ready to be integrated with MLFLow tracking.

Urbana-Champaign Area

Research on creating a reliable extraction and indexing framework for chemical diagrams from scientific documents using graph-based molecular representations to be used in a reaction tracker tool.

Rochester, New York Area

Real-time Monocular Absolute Depth on a smartphone using CoreML: Developed anovel method for estimating absolute depth from an RGB Image using a deep learningtechniques for efficient indoor localization. (NDA - Patent application filed).Collecting training data, annotating data and training a deep learning model to detectand localize common office objects including printers in realtime.Met all requirements including optimizing the trained model and deploying it on anandroid device (TF-Lite) for real-time inference at 15 FPS.

Rochester, New York, United States

Trained and optimized Tensorflow models for inference on TI's AM57 chip using TVM (TI's deep-learning accelerator).

Kolkata Area, India

Doctor Of Philosophy - Phd, Computer And Information Sciences

PhD student in the DPRL lab Working on detection and parsing of math and chemical expressions for extraction, construction, indexing.

Master Of Science - Ms, Computer Engineering, 3.86

Bachelor Of Technology - Btech, Electrical, Electronics And Communications Engineering