• Hi!
    I'm Divyesh

    Senior Research Assistant at SUNY Research Foundation

    Download Resume

  • I am a
    Machine Learning Engineer

    Specializing in Multi-modal Active Speaker Detection, and Large Language Models

    Contact Me

About Myself!

Hi Visitor!

In one line, you could call me a classical fan of the A.I. Sci-fi TV show: Person Of Interest. During my undergrad years, I came across the show on Artificial Intelligence, that was way ahead of its time, and I was awstuck by the forward-thinking depiction of AI. Luckily enough I got the opportunity to work on projects related to Data Science in my early corporate years (where I got the Best Graduate Engineering Trainee among 270 other new joiners), and 6 years later I am pursuing a MS degree in Artificial Intelligence where I am grateful to contribute my small part in leveraging A.I to simplify people lives, and making it better.

Cut to the chase, I am a Machine Learning Engineer with over 5 years of full-time experience in the automotive, finance, and healthcare industry . Currently I work as a Graduate Research Assistant at the SUNY Research Foundation focusing on Deep Learning research.

I specialize in developing Machine Learning pipelines using state-of-the-art architectures like BERT, RAG, Whisper, CVAE, and LLM models such as LLaMA and GPT. Additionally, I am proficient in traditional data science models, time series forecasting, and statistical analysis, including hypothesis testing and A/B analysis.

Large Language Models, Prompt Engineering, Active Speaker Detection, and Pose Estimation

Industries: Credit Card & Banking Sector, Automobile, Retail and Healthcare

Traditional ML, A/B testing, Predictive Analytics, and Time-Series Forecasting.

Big Data, ETL, Cloud Computing, Agile, Kanban and CI/CD

Math Inspiration from POI (Person of Interest):
The number pi stores every numerical passcode, every single phone number in the world, and certainly my DOB at the 48,696,598th position!
Now you know my Date of birth :)

Watch the short clip here!
Years of Full Time Work Experience
Professional Projects
Organizations
Times Best Employee
What I am up to right now!

Recent Projects

HTML5 Bootstrap Template by colorlib.com
May - July 2024 | Computer Vision and RAG based Comprehensive Car Repair Product | 4

Automobile Inspector

AI-driven Full-Stack web application integrating object detection (ResNet) and image segmentation (Deformable Convolution Network) for comprehensive car damage detection & cost estimation in just 5 clicks. It also integrates a RAG based conversation chatbot for repair agent matching & online support

HTML5 Bootstrap Template by colorlib.com
May 2024 - Present | Computational Linguistics, Prompt Engineering, and Data Annotation | 4

Noun Verb Syntax Observation Network

NVson is a comprehensive automated data annotation pipeline leveraging LLM, and RoBERTa based pos tagger model designed to automate Part-of-Speech (POS) tagging for previously uncategorized tags.

HTML5 Bootstrap Template by colorlib.com
July 2024 - Present | Near Real-time Multi-Modal Active Speaker Localization and Transciption | 4

Audio-Visual Active Tracking and Annotation Rendering

AVATAR processes live camera feed to detect, tag & track multiple faces using S3FD, InsightFace & SORT respectively, evaluates lip-sync scores using SyncNet/ TalkNet model, & finally annotates active speaker in real-time using a Producer-Consumer Multi-Threading Architecture.

HTML5 Bootstrap Template by colorlib.com
October - December 2024 | Reinforcement Learning | 4

Pong Agent

This project implements RL Deep Q-Network (DQN), DDQN and PPO techniques to win the Atari Pong game against a computer.

HTML5 Bootstrap Template by colorlib.com
January - April 2024 | Binaural Beats Latent Space Modelling | 4

NeuroBeats

Deep learning Latent Space exploration (Conditional Variational Encoder) of EEG-processed brain topographical maps, uncovering the impact of binaural beats on brain activity.

Experience

Full Time Work Experience

Research Assistant at The Research Foundation SUNY April 2024 ‑ Present

Modelled and trained a 25M parameterActive Speaker Detection model on Oxford Voxceleb, and Google AVA, AVSpeech dataset consisting of 2 Million videos to generate Lip Sync Scores. The model performs with a 0.89 F1 score. Major improvement is this is the first near Realtime Active Speaker Localization model that works on live camera feed with a latency of 1.2 seconds. All other pipelines works only on offline videos.

Data Science Lead Assistant Manager at EXL Analytics November 2021 ‑ June 2023

Implemented data quality controls for Credit & Fraud risk models. Developed strategies to adjust credit lines and manage adverse actions based on transactional behaviors. Reduced delinquent losses by 5% and manual verification efforts by 30%.

Analytics Deputy Manager at Suzuki Motors India Limited July 2017 ‑ November 2021

Designed and introduced a forecasting model by integrating ABC and XYZ analysis to provide live insights on SKU availability, safety stock, & reorder points for over 4,000 parts. Increased inventory turnover ratio from 3.2 to 4.7, leading to annual cost savings of $120K by reducing wastage.

Software Verification & Validation Intern at Alstom Transport Jan 2017 ‑ June 2017

Performed Black Box testing on a SIL 4 Automatic Train Supervision (ATS) software system (Kochi Metro) as per CENELEC 50128 standards. Kochi Metro is India's First Communication Based Train Control (CBTC system).

Education

Full Time Degrees

University at Buffalo, The State University of New York, USA

GPA: 3.8 / 4.0

August 2023 – May 2025

Relevant Courses: Machine Learning, Computer Vision & Image Processing, Pattern Recognition, Reinforcement Learning, Advanced Analysis of Algorithm & Design, Data Intensive Computing, Information Retrieval

Thapar University, Patiala, India

CGPA: 8.23 / 10

July 2013 – June 2017

Relevant Courses: Data Structures, Linear Algebra, Calculus, Signal and System, Numerical Maths, Digital Electronics, Analog Circuits

Certifications

Artificial Intelligence

Machine Learning

Introduction to modern machine learning and associated maths, including supervised learning (multiple linear regression, logistic regression, neural networks, and decision trees), unsupervised learning (clustering, dimensionality reduction, recommender systems), and some of the best practices used in Silicon Valley for artificial intelligence and machine learning innovation (evaluating and tuning models, taking a data-centric approach to improving performance, etc.)

Neural Networks and Deep Learning

Build, train, and apply fully connected deep neural networks; implement efficient (vectorized) neural networks; identify key parameters in a neural network’s architecture; and apply deep learning to your own applications.

Sequence Models

Speech recognition, music synthesis, chatbots, machine translation, natural language processing (NLP)

Improving Deep Neural Networks

Hyperparameter Tuning, Regularization, and Optimization.

Structuring Machine Learning Project

Diagnose errors in a machine learning system; prioritize strategies for reducing errors; understand complex ML settings, such as mismatched training/test sets, and comparing to and/or surpassing human-level performance; and apply end-to-end learning, transfer learning, and multi-task learning.

Convolution Neural Networks

Comprehensive understanding of Convolutional Neural Networks, including architecture design, training techniques, and application areas.
Get in Touch

Contact