Hi! Welcome to my profile. I'm

GAUTAM VR




A Little about me

  • Age: 28
  • Location: New York, NY, US

  • Currently, An AI Software Engineer at a Series-A funded startup Instalily.ai, working on building AI powered solutions for the healthcare and construction industries, generating incremental revenue of around $47M.

  • I have graduated from Columnia University, New York with a Master's degree specializing in AI and ML. Along with my rich course work in the subjects of Applied ML, Deep Learning and Cloud Computing, I've also researched in the field of Vector Database and RAGs.

  • I've been as a software engineer with 5+ years of industry experience, previously worked in Tech organizations like Intuit and Philips, where I've not only contributed to the growth of the organization but also to my personal growth in the fields of software engineering and AI applications.

  • An AI Enthusiast certified by Tensorflow, who loves building exciting Deep learning models and has a strong hands-on knowledge in Deep learning and Computer Vision through various projects, internships, and publishing IEEE research paper.

  • AWS Certified cloud developer who has a sound hands-on experience in building cloud native applications using Amazon Web Services & Google Cloud Platform.


    • Few other Interests..

  • I love traveling to many different places across the country, to get along with people with different cultures and to gather lots of cool travel stories.

  • I’d always be up for a game of football, be it on turf or on PC/Playstation. I also do love watching football along with other sports.

  • Reading books and gardening are few other activities which I've grown huge interest towards lately.

    • Where I've worked at

    Professional Experience

    Instalily.ai

    New York, NY, USA

    AI Software Engineer

    May 2025 - Present
    • ML Modelling at Ensemble Health Partners - Built ML models for COB denial prediction with 91% recall that resulted in $1.2M cost reduction.
    • Laid the infrastructure for modelling & scoring pipelines using AzureML, leading to seamless deployment for 7 clients.
    • Cosailor AI at SRS Distribution - Developed and owned 3 vital AI modules used in the Cosailor - AI powered sales agent that generated incremental revenue of $47M for SRS construction distributor.
    • Built modules using LLMs for "TODO tasks generation from notes", "email generation with personality index" - focused toward sales representatives with 87% adoption rate.
    • MCP Platform - Integrated RBAC OAuth2.1 authorization framework internally for enhancing security within AI modules.

    Intuit

    Bangalore, India

    Software Engineer 2

    Oct 2022 - Sept 2024
    • Worked in core development of ProSeries TY22 & TY23, Professional Tax software products for US which has revenue over 170 million USD.
    • Successfully released 11+ releases for the product, having 50k+ customers in US, using Jenkins and analysed various data in Splunk dashboards.
    • Filed patent and honoured for incorporating Intuit Assist (GenAI platform) into ProSeries TY23, using paved path services to reduce 60% of preparation time for tax preparers.

    Philips Healthcare

    Bangalore, India

    Software Engineer 2

    Oct 2021 - Present
    • Carried out the ML Clinical pipeline development, by modelling ML models with data ingested from Clinical Data Lake (CDL).
    • Led the end-end development cycle of 3 major features from high level system design to development & code reviews.
    • Awarded with the best innovative project for prototyping Medical image segmentation (PyTorch) to identify tumour anomalies in medical images by incorporating transfer learning & distillation.

    Software Engineer 1

    June 2019 - Sept 2021
    • Owned & delivered 4 major components using object-oriented programming along with design diagrams and requirements.
    • Implemented 6 features in C# while adhering to SOLID principles using BDD (SpecFlow), ensuring code quality TICS > 95%.
    • Fixed 3 major Change Requests to address critical customer feedback & investigated 18+ P1,P2 defects in MRI R10 release.

    Internships & Research

    Columbia University

    New York, NY, USA

    Research Assistantship

    Jan 2025 - May 2025
    • Worked with extensive research with memory optimization in VectorDB's system design and RAG pipeline for code generation.
    • Analysed the Starling and DiskANN memory optimization techniques to improve the in-memory cache of the VectorDB's efficient retrieval.
    • Compared the chunking techniques for the RAG code generation pipeline.

    Teaching Assistant

    Oct 2024 - May 2025
      Supervised as the lead TA for the following courses, with the responsibility of conducting weekly labs, grading assignments and projects, and helping students with their doubts :

    • NextGen Cloud Computing (ELEN6770 - Fall'24) - Covered topics to design and manage complex cloud systems using advanced tools like Kubernetes, serverless functions, and multi-cloud architectures.
    • ML in Networking (ELEN6772 - Spring'25) - Appling machine learning techniques to optimize network performance, security, and traffic management in modern networking systems.

    AIrLitz

    New York, NY, USA

    Cloud Engineer Intern

    Sept 2024 - Jan 2025
    • Implemented the end-to-end image ingestion workflow with AI object detection by fine-tuning CLIP with 77% accuracy using AWS step func-tions.
    • Performed detailed analysis of cost vs. performance for AWS services, which improved the pipeline speed by 300% and cut down costs by 80%.

    Philips Healthcare

    Bangalore, India

    Software Engineer Intern

    Jan 2019 - June 2021
    • Helped in creating a dev framework to independently launch IAP plugins by mocking BU specific components

    Nokia Solutions and Networks

    Chennai, India

    Data Analytics Intern

    May 2018 - June 2018
    • Worked for product engineering with Data Analytics & Machine learning and reduced 72% of the defective modules in production line.

    A few projects that I've built

    Patient Monitoring System

    • Rules based Alerting System, a desktop application that provides all the relevant services for hospital staff to remotely monitor all patients and their vitals in the ICU.
    • Developed using C# in .NET Framework with appropriate design patterns and using MVVM architecture for UI Layer.
    • Created RESTful API endpoints using WCF to for various functionalities in alerting and monitoring.
    • Integrated with gating for code metrics, unit tests using NUnit.

    Chatbot - Healthcare Assistance

    • An application that assists the end user in finding a suitable patient monitor/integrated solutions based on all products available along with the list of features of the product or a purchase suggestion based on specific features that the user prefers.
    • Developed in C# with .NET framework integrated with a CI pipeline that builds, runs tests and generates report.
    • Uses MS SQL database to store the pre loaded messages and the information of the products and solutions that are offered.

    AWS Web Application

    • A 3-tiered Cloud native Web application using various AWS services along with a CI/CD pipeline to flawlessly integrate and deploy the updates
    • Created a custom VPC with subnets in a region, across multiple availability zones to deploy the application with high availability and more security.
    • Used node.js with express to create REST APIs with authenticated endpoints using JWT and hosted on EC2 instances under an ALB with autoscaling enabled.
    • RDS is used as the main database server with MySQL provisioned with read replicas. Other AWS services like S3, SES, CodeDeploy, Lambda, CloudWatch.
    • Web application is integrated with CI/CD pipeline which runs tests and deploys the application using CodeDeploy to the servers along with other AWS services like S3, SES, Lambda, CloudWatch.
    • Created the infrastructure with Cloudformation (IaC) for reuse of stacks.

    Document Analyser - GCP

    • An AI system using Google Cloud Platform's DocumentAI service which takes a PDF document (bills/receipts) and extracts the relevant data from it and stores them in Key-Value pairs for further automated analysis.
    • Training data is pulled from the BigQuery public database which has PDF files in Cloud Storage of patents from the United States and European Union.
    • A training pipeline which formats the training data and uses AutoML to build Image Classification, Entity Extraction, Text Classification, and Object Detection models.
    • Results are stored in BigQuery tables to produce results with relevant fields.

    AWS - Music Streaming App (Serverless)

    • An AWS Server-less music streaming application with stack created using AWS CDK (Cloud Development Kit) - in python.
    • Created using the AWS resources that are created as part of the CDK stack
    • Resources - API Gateway, Lambda, Kinesis Data stream, DynamoDB, S3 bucket

    SkimLit

    • A Natural language Processing (NLP) model that skims a research paper and provides a categorical summary of the same.
    • Analyzed various RNN models such as GRU, LSTM and also with CONV1D as RNN
    • Achieved around 82% accuracy over the test set

    Space-X Landing Analysis

    • An end-to-end data science project in python that predicts whether the landing outcome of the SpaceX's Falcon 9 first stage will be successful or not.
    • Data Collection with ETL is done from SpaceX's APIs and wikipedia using BeautifulSoup and requests api.
    • Exploratory Data Analysis (EDA) through SQL and visualization in plotly
    • Interactive Visual Analytics using Folium and Plotly Dash
    • Predictive analysis using various ML classification models (KNN, Decision tree Classifier, Logistic Regression and SVM) using GridSearchCV to find the optimal hyper-parameters.

    FoodVision

    • A deep learning Computer Vision model that identifies an image of any food item (over 101 food categories)
    • Uses EfficientB0 model architecture which has over 5.3 Million parameters
    • Trained the model using transfer learning, by fine tuning the last 10 layers over a dataset of 101000 images (1000 images in each category)
    • Achieved around 89% accuracy over the test set.

    Photoscope - Desktop Application

    • An image search desktop application, similar to Google images, where you can search and view images from Flickr on Desktop.
    • Created using .NET Framework and WPF to create the client UI and the functionalities.
    • Developed using MVVM framework for UI - Service layer interaction, through appropriate design patterns.
    • Service layer uses the Flickr API to search for images in the web using various search parameters and displays a feed with photos and it's comments
    • Functionalities such as Caches, Logging, Multi-threading, memory-management and pipeline for seemless developement is also Incorporated with this developement.

    Snake And Ladder Game - WPF

    • A simple snake and ladder game developed using WPF
    • Created in .NET framework with C# including timers, dynamic bindings, etc.

    Technology stacks that I've worked on





    Digital Badges and Certifications

    Please reach out to me

    I love to build many interesting projects on my own time, and I'll always be up to colloborate on building new stuffs and exploring new technologies, So please feel free to contact me if you are looking to team up or even to have some nice coffee chat.

    Call or Whatsapp me at :

    +1 (646) 238-7264

    Location:

    New York, NY, USA

    Loading
    Your message has been sent. Thank you!