Smit Patel Email & Phone Number

United States

Pittsburgh, Pennsylvania, United States

• Assisted Prof. Hong Shen, for her class on Critical AI Studies for Public Policy, as the only TA for a class of 25+ students. Responsibilities include:
• Resolving student queries during and beyond office hours
• Aiding the Professor in framing examination questions
• Managing operational and daily classroom logistics
• Grading, facilitating and monitoring discussions followed by the final Policy Memo

Pittsburgh, Pennsylvania, United States

Researched with Prof. Rema Padman in iLAB on Healthcare Analytics

Bengaluru, Karnataka, India

• Leading the C++ Core Technology team in the Modular Interiors division generating $1M+ revenue
• Designed Moller-Trumbore raycaster that boosted object detection efficiency by 50% & reduced memory footprint by 35%
• Built Affine Transformation Matrix system to power Scale-Rotation-Translation operations on OpenGL shapes in real-time
• Built a robust CPQ Pricing framework facilitating precise, real-time price computation for all SKU configurations
• Developed and deployed key business modules like Panel Cutlist, Fulfillment Barcode Generation & Bill of Quantities

Palo Alto, California, United States

• Developed sentiment analyzer using DistilBERT-based ABSA & analyzed 5000+ customer feedback
• Integrated multi-threaded video recorder using OpenCV, ffmpeg & H.264 compression algorithm

Singapore

• Researched at the Computational Intelligence Lab at NTU Singapore
• Attributed Twitter sentiment analysis to the prediction of real-time Bitcoin trading volume
• Fine-tuned FinBERT and VADER ensemble for Twitter sentiment detection, achieved 0.832 F1 score
• Employed Information Gain for feature weights and CKG-based Concept Parsing for mining relevant data
• Achieved remarkable P-value of 3.89 ∗ 10−15 and LSTM loss of 0.0094 on a large dataset of 2M+ tweets
• Proceedings published in A* International Conference on Information Systems 2022 as a conference paper

Indore, Madhya Pradesh, India

• Developed a novel Vision Transformer (ViT) Siamese model fine-tuned for Fingerprint Authentication
• Enhanced self-attention mechanism by 43% using Learnable Temperature Scaling & Diagonal Masking
• Implemented Shifted Patch Tokenization for increasing locality inductive bias and achieved TPR of 0.817
• Research proceedings published as a conference paper in ICCPN 2022, Thailand

Auckland, New Zealand

• Designed a holistic Code Clone detection pipeline using multiple Siamese Neural Network architectures
• Obtained P, R and F1 scores of 0.902, 0.933, and 0.917 on OJClones dataset with Intermediate Merge Siamese Neural Network
• Incorporated Abstract Syntax Tree and Control Flow Graphs for holistic source code embedding of dataset
• Achieved a 23-fold increase in training time efficiency compared to state-of-the-art approaches
• Research Proceedings published as a first-authored full conference paper in ICTSS 2022

Hyderabad, Telangana, India

• My responsibilities primarily involved improving PostTrade Pricing service with SpringBoot, Mockito and migration to Microservices architecture.
• Performed migration of the backend Java service and DAO layer and optimized the business logic increasing the service efficiency by 12%

Mumbai, Maharashtra, India

• Built a Paraphrase Generator using Seq2Seq model with Encoder (GRU), Decoder (LSTM) and SharedPair-Wise Discriminator for learning semantic sentence embeddings
• Integrated lexical semantic analyzer using Knowledge Graph-based taxonomy tree and Resnik Similarity
• Incorporated tense transformation and achieved remarkable BLEU and ROUGE-L scores of 51.5 and 54.7

Mumbai, Maharashtra, India

• My responsibilities included detecting and eliminating inaccurate data by implementing bounce rate filters using Javascript
• Built dashboards and published lead qualification and product landing pages having an engagement of 850+ users and 2.1k+ sessions

Master Of Science - Ms, Artificial Intelligence And Innovation

Bachelor'S Degree, Electrical Engineering, 9.01/10

Minors: Humanities and Social Sciences

Science, 94%

High School, 96.33%