EDUCATION

WORK EXPERIENCE

• Software Engineer II, Amazon (AWS) / Amazon Alexa

  April 2022 - Present · Bengaluru, KA

  Tech: Java, Python, TypeScript, Express.js, Node.js, Next.js, AWS cloud stack
  
  → Config driven autonomous agentic system for launching Fee type to a new marketplace. Launching a fee to a new marketplace used to take 26 weeks of SDE effort; with autonomous agents, this was reduced to ~1 week.
  → Inventor in Patent in review for Alexa+ conversational bug reporting or feedback mechanism: Feedback Expert.
  → Retrieval-augmented generation (RAG) system for Alexa Feedback processing to support 1850 TPS traffic.
  → Designed and developed frontend and backend services at AWS re:Post. Worked on SEO and A11Y which resulted in performance improvement of 86% of page hits based on Google Analytics.
  → Integrated Content Curation model to moderate content in re:Post.
  → Multi-language feature developed for repost.aws, increasing customer usage by 4 folds.

• Senior Software Engineer II, Sophos MTR

  June 2021 - April 2022 · Montreal, QC (Remote)

  Tech: Python, Gunicorn, Golang, Terraform, Bash, Kafka, Kinesis, SNS, SQS, Parquet
  Tools: Spring Boot, Maven, ELK stack, Kubernetes, Jenkins, Pytest, AWS
  
  → Designed and developed consumers/producers to store streaming endpoint/firewall data to data lake.
  → Setup telemetry and alerts for monitoring and performance analysis.
  → Reduced EC2 instances required by 8x (e.g. 128 to 16 in us-west-2) by migrating consumers from Python to Golang.
  → Deep learning model for threat detection in packets captured via pf_ring by converting them into spiral images.

• AI Developer II, Intact

  August 2020 - June 2021 · Toronto, ON

  Tech: Python, Rule Engine, Consumer, Drools, Java, MongoDB, Confluent Kafka
  Tools: Spring Boot, Maven, ELK stack, Kubernetes, Git CI, Pytest, AWS
  
  → Architecture, Design, Build, Implement and Test claims document classification consumer component.
  → ML Model for insurance claim classification using an ensemble of NLP and image based deep learning model.
  → Developed A/B testing model pipeline for different model deployments based on language and region.
  → Optimized ML model deployment pipeline to enable push button model update.
  → Setup Logging pipeline to separate user logs with application logs for D&A dashboards.
  → Conducted stakeholder demos, internal training and coaching for code quality.

• Research Software Developer, Via Science

  January 2019 - June 2020 · Montreal, QC

  Tech: Python, PostgreSQL, SQLAlchemy
  Tools: Django, Tornado, Pytest, Asyncio, Git, AWS, Conda, Redis, Docker, RabbitMQ, Celery
  
  → Designed and developed RESTful web services and Stateless servers following “The 12 Factor App” methodology.
  → Developed scoring algorithm as labels for transformers and used it to create multiclass ML model.
  → Enabled User Acceptance Testing by revising the conda package to enable product usage.
  → Designed and developed pipeline for integrating ML models in the product for customer usage.
  → Wrote scripts for automated integration tests and Behaviour Driven Development tests.

• Research Assistant, McGill University

  January 2017 - August 2018 · Montreal, QC

  Tech: Python, Angular, Node.js, PostgreSQL
  Tools: Phylo, OpenPhylo, Citizen Science, AWS, Docker, RabbitMQ
  
  → Proposed a new framework for game Phylo after analysis of data collected since 2010. Paper published in HCOMP.
  → Enhanced difficulty prediction of puzzles by ~12% using Neural Networks. Paper published in AAAI.
  → End-to-end development of OpenPhylo reducing manual effort of extracting Phylo puzzles to 0.
  → Integrated Teaching Portal to OpenPhylo increasing players count by 15%.
  → Phylo is deployed in Science and Technology Museum, Ottawa.

• Teaching Assistant, McGill University

  September 2017 - December 2018 · Montreal, QC

  Tech: Java, UMLet, MEAN stack
  
  → Designed an assignment (with solution) requiring students to do Test Driven Development (TDD).
  → Provided comprehensive tutorials on 12-Factor-App.
  → Answered course related questions from students on the course discussion board.

• Programmer Analyst, Cognizant

  June 2014 - June 2016 · Chennai, IN

  Tech: Java, J2EE, Hibernate, MySQL
  
  → Architectured, designed and implemented extraction of variables from Calligo scripts to increase productivity by ~13%. Digital Superstar for 2 quarters.
  → Automated the process of maintaining bugs related to similar issues and assigning them to the same person.
  → Made way for the adoption of xDashboard stack by client via a successful PoC.

PUBLICATIONS

• A storified citizen science computer game to teach and contribute to genomic data analysis

  Co-author · Genomics · July 2022

• Lessons from an Online Massive Genomics Computer Game

  First-author · AAAI · September 2017

  Identify patterns that contribute to success and pitfalls in a Citizen Science product. Show how Phylo has been used to improve a reference alignment of Ebola virus sequences.

• A HCI platform for multi-scale genome analysis

  First-author · HCOMP · August 2017

  A citizen science framework for collective curation genomic annotation at multiple levels of ancestry.

• Chaos Based Cryptosystem for Images

  First-author · IJEDR · Fall 2013

• Debit card security based on chaos function and QR code

  First-author · IEEE · Winter 2014

PROJECTS

• OpenPhylo (Thesis)

  Designed an improved crowd-computing platform using CNN-based MSA region identification, difficulty prediction, player-puzzle routing, and enhanced feedback mechanisms to solve the multiple sequence alignment problem.

• Phylo

  Developed ML models and RESTful APIs (Node, Express, MongoDB) for a citizen-science game that crowdsources solutions to the Multiple Sequence Alignment problem.

• Stackoverflow #tag prediction

  Built an NLP model using multilabel classification to predict tags for StackOverflow posts.

• Mini Project: Lessons from a letter based language classifier

  Compared Naïve Bayes, Logistic Regression, k-NN, and SVM for letter-based language classification using centroid-based and tf-idf approaches.

• A New Chinese Dialogue Corpus Extracted from Baidu Tieba

  Introduced a Chinese dialogue corpus designed for training goal-driven end-to-end dialogue systems.

• QCidRoster

  Built a tool to track change requests and intelligently assign defects to the most suitable developers based on requirement document knowledge.

• Teaching Portal

  Built a portal for instructors to create MSA assignments, finding that domain-knowledgeable students outperform casual players on difficult puzzles.

• ICLR Reproducibility Challenge

  Investigated the reproducibility of the ICLR 2018 paper “Convolving DNA Using Two-Dimensional Hilbert Curve Representations.”

• Concern-Oriented Reuse (CORE) for Hibernate Framework

  Reverse-engineered Hibernate ORM to create a reusable CORE file by weaving Reusable Aspect Models (RAM) and generating source code from UML 2.0 sequence diagrams.

• Reverse Engineering of WSO2 IoTServer

  Conducted a comprehensive architectural reverse-engineering study of the WSO2 IoT device-management-server framework, analyzing its static and dynamic components.

Certifications

• AWS Certified (Credential ID: 5FNBKH92D2VQQBK0)