VENKATAKRISHNAN RANGANATHAN

AI Engineer

Contact Information

Education

Master of Science (M.S), Artificial Intelligence

Northwestern University Evanston, Illinois, December 2023

GPA: 3.87 / 4.00

Relevant Coursework: Deep Learning, NLP, Logic in AI, Python, Data Science, Approximation Algorithms, Generative Deep Models, Computer Vision

Bachelor of Technology (B. Tech), Information and Communication Technology

SASTRA University Thanjavur, India, 2010

GPA: 7.47 / 10.00

Skills

Recent Projects

Analyzing the impact of fine-tuning for summarization of longer sequences in PEGASUS, 2023

  • Investigated the performance of fine-tuning using the Multi-News dataset for varying input lengths, specifically focusing on the impact of exceeding the maximum input length.
  • Measured advancements in summarization systems and gained a comprehensive understanding of the effects of fine-tuning on lengthier sequences.

Legal Case Consolidation using Unsupervised Methods, 2023

  • The objective involved grouping legal cases that had traversed different hierarchical levels within the courts.
  • Employed unsupervised methods such as TF-IDF, Sentence Transformers, and Legal-NER while utilizing vectorization and semantic similarity search techniques like Cosine Similarity and FAISS.
  • Achieved 90% accuracy on grouping cases based on tests conducted on data from LEXIS.

Fake Bio Detection using FFN, LSTM and Transformers, 2023

  • To detect fake biographies generated by a language model from a mixed corpus of real and fake biographies.
  • Adapted BERT model with 110 million parameters to train it for binary classification tasks.
  • Attained an accuracy rate of nearly 94% in the detection of fabricated biographies.

Unsupervised Coloring Book Image Generator with CycleGAN, 2023

  • The goal was to create a line drawing image from a natural photograph that could be utilized for coloring purposes.
  • Trained an image generator using GAN with unpaired dataset of real-world images and curated coloring book-style images.
  • Successfully obtained high-quality image outputs, consisting of line drawings depicting objects from the original natural photographs.

Real-time Detection and Correction System for Improved Sitting Posture, 2023

  • To provide real-time feedback to computer users about their sitting posture and suggest corrective measures without intrusive devices.
  • Implemented self-defined calibration and threshold values for accurate slouching detection using MediaPipe and OpenCV.
  • Captured the posture of an average adult and built a notification system to deliver real-time feedback.

Disease Classification based on low-resource Bean Leaf images dataset, 2023

  • The objective was to categorize leaf images with diseases, despite having only a limited dataset of a few hundred examples.
  • Constructed a traditional CNN model and an auto-encoder using healthy plant leaves from diverse species, and employed the encoder component to perform classification tasks on the bean dataset.
  • Achieved over 90% accuracy on identifying diseased bean leaves using the Auto-encoder model.

Detecting Aortic Stenosis from a single lead (Lead II) ECG using CNN, 2022

  • The task focused on detecting Aortic Stenosis from a single lead (Lead II) electrocardiogram (ECG).
  • Denoising and feature extraction were performed on raw ECG signals to train a Random Forest classifier for disease detection. Additionally, a CNN was constructed by directly inputting the denoised signal.
  • The CNN model yielded results with an accuracy of 88%.

Professional Experience

Research Assistant, Institute for Policy Research, Northwestern University

Jun 2023 - Present

Constructed machine learning models for extracting information from tribal legal documents, U.S. case law documents, and court case transcripts through summarization and sematic parsing.

Utilized character level NLP techniques to help identify offensive Native American high school mascot and street names.

Conducted a comprehensive analysis and modeling of the complete historical record of cases ever prosecuted in the United States, like the U.S. case law history network.

Collaborated with technical experts to overcome challenges in network analysis and visualization.

Data Scientist/Data Engineer at Tata Consultancy Services Ltd. (2019-2022)

  • Developed a comprehensive data visualization portal for a social engagement platform (Campus Commune) using Grafana to help the recruitment team spot talent and make strategic decisions on engagement models.
  • Selecting features, building, and optimizing classifiers using SciKit Learn and PyTorch.
  • Enhancing data collection procedures by manual scrutiny to include relevant information for building analytic systems.
  • Led two key human resource focused projects: one focusing on predicting and placing candidates by their field of expertise using Python and the other a joining predictor using ML Models.
  • Adapted Unsupervised K-means clustering for classifying comments received on company’s engagement portal based on relevancy, uniqueness and value-add.
  • Collaborated on multiple ad-hoc projects pertaining to recruitment, to analyze, process, cleanse, visualize and verify the integrity of such data using Pandas and presented the results for upper management for decision making.

Cloud Architect and DevOps Specialist (2015-2022)

  • Designed and implemented various Linux based microservices like search engine, cache store etc. and deployed Python Machine learning APIs using Docker containers for a social learning/engagement initiative by Human Resources (HR) Department.
  • Led the team to implement Automation of deployment and testing using configuration management, server orchestration and continuous delivery/integration (Ansible, Gitlab and Capistrano).
  • Managed end to end implementation and designed cloud-based solutions for various public facing online learning applications for Corporate Social Responsibility (CSR) department.
  • Prepared Capacity and Budget planning of the Cloud infrastructure which hosted multiple applications for HR and CSR departments.
  • Configured scalable Relational Database system for various applications (MySQL Database scaling for 10s of thousands of concurrent users. load balancers, Memcached, Redis).
  • Conducted regular Vulnerability Assessment, Penetration Testing and Patch management for securing various environments for multiple applications. Reviewing and debugging Error logs, Network Flow Logs, and guiding the team for fixing the issues found on Application components.
  • Optimized Cloud usage and minimized the monthly costs by Implementing dynamic scaling using AWS auto-scaling via Infrastructure as Code tool (Terraform).
  • Collaborated with multiple teams for migrating data and application hosting from Amazon Web Services to Google Cloud Platforms.
  • Fine-tuned network components in Google Cloud Platform (GCP) and Amazon Web Services (AWS) via Load testing (JMeter) to handle high traffic on all the above-mentioned applications.

Software Developer (2010-2015)

  • Co-led a team to ideate and implement an Online coding module, a patented gamified learning platform that improves qualitative written/spoken English, Knowledge Currency system for HR Department’s student engagement portal using Ruby on Rails and MySQL.
  • Spearheaded a team of designers, developers for ideation and implementation of a gamified Peer to Peer Quizzing/Learning System.
  • Architected and implemented an online module for conducting various online contests for college students across the world.

Additional Information

Implemented an end-to-end autoscaling infrastructure for handling 100000s of concurrent users on IaaS hosted Web Applications

Finalist at an inter-company innovation competition for the Joining Predictor using Machine Learning Models 2019