Kavish Gandhi

Software Engineer

Greater Toronto Area, Canada5 yrs 3 mos experience
Most Likely To SwitchAI ML Practitioner

Key Highlights

  • Optimized performance for Alexa App routing.
  • Developed CV algorithms for accessibility.
  • Led team to improve ML model accuracy.
Stackforce AI infers this person is a skilled AI/ML engineer with a focus on performance optimization and accessibility solutions.

Contact

Skills

Core Skills

Machine LearningDistributed TrainingSoftware DevelopmentApi DevelopmentComputer VisionResearchEdge AiProject Management

Other Skills

Python (Programming Language)ProgrammingResearch SkillsFull-Stack DevelopmentAngularpandasFastAPINumPyLeadershipBack-End Web DevelopmentData ScienceBlockchainTensorFlowBitBakeTFlite

About

I'm an AI enthusiast with a range of software development experience, passionate about building impactful solutions that solve real-world problems. Currently, I am working on AWS Neuron SDK optimizing distributed ML training and inferencing Here are some highlights of my journey: - Building AI for accessibility: I'm on a mission to make information more accessible to everyone. Prior to Amazon, I developed CV algorithms to help visually impaired individuals understand and interact with graphs. - Optimizing performance: I'm driven by efficiency and performance. At amazon, I optimized routing between pages/view on the Alexa App (improvement of ~10ms). At MediaTek, I contributed to optimizing source code which led to a 50% speed improvement for ONNXRuntime tasks on their SoCs. At Merkle, my PySpark report generation application slashed ~9 hours of manual work and manpower required (from 4 to 1), thereby saving 75% of time and efforts - Leading and collaborating: I thrive in both individual and collaborative environments. While at Ugam, I led a team in developing an ML model that achieved 69% accuracy in multi-class binary label classification - a significant improvement over previous models. - A passion for learning: I'm always seeking new challenges and opportunities to learn. In my personal projects, I've explored NLP applications like generating code descriptions and developing creative ML project ideas. Connect with me if you're interested in: - AI and its applications across various domains - Full-stack development and building robust software - Optimizing performance and achieving efficiency - Collaborating on impactful projects and solving real-world problems Let's chat! P.S. I'm passionate about sharing my knowledge and experiences. Feel free to reach out if you have any questions or are interested in learning more about my work.

Experience

5 yrs 3 mos
Total Experience
1 yr 4 mos
Average Tenure
2 yrs
Current Experience

Amazon web services (aws)

2 roles

Software Engineer II - AI/ML

Promoted

Oct 2025Present · 6 mos

Software Engineer I - AI/ML

Apr 2024Oct 2025 · 1 yr 6 mos

  • Part of Neuron Distributed Training team
Python (Programming Language)Distributed TrainingMachine Learning

Amazon

Software Engineer

Aug 2023May 2024 · 9 mos · On-site

  • Design, implement, test and productize the ESlint rules to identify specific words to enforce the governance to detect the release halters on 170+ packages from 60+ partner teams
  • Optimized routing between pages/views on the Alexa App (improvement of ~10ms)
  • Single-handedly detected an error that caused the Alexa App to crash and helped the required team to mitigate the error and have been active on-call from 4th week of joining
  • Contributed to the design of monitoring dashboards for teams' specific metrics, enabling early anomaly detection and timely remediation, minimizing customer impact
ProgrammingAPI DevelopmentSoftware Development

Georgia institute of technology

Research Assistant

Feb 2023Aug 2023 · 6 mos · Atlanta, Georgia, United States · Remote

  • Research & Develop CV algorithms for graph ingestion, making graph images more accessible to the visually impaired – extract data, metadata, color, title, legend, etc. from the static image of a graph
  • Built a multiclass classification model to classify graphs and developed a web app which enables users to use the ML and CV models
Research SkillsComputer VisionResearch

Mediatek

Edge AI

May 2022Aug 2022 · 3 mos · United States

  • Part of Edge AI Team
  • Contributed to proprietary source code helping build delegates enabling customers achieve faster results on MediaTek SoC devices (~50% faster for ONNXruntime, when compared to CPUs)
  • Developed scripts to automate time consuming manual tasks, saving time and efforts
  • Built visualisation tools helping customers do better analysis of models
ResearchPython (Programming Language)Edge AI

Ugam

Research Analyst (R&D Team)

Jul 2019Aug 2021 · 2 yrs 1 mo · Mumbai, Maharashtra, India

  • Part of the R&D team and was responsible for coding backend services to cater to the client-facing teams.
  • Researched and implemented ML+CV models for classification.
  • Led the newly joined team at Ugam in creating PySpark ML classification model with ~69% accuracy which was an improvement of ~3% over the previous model used
  • Developed a PySpark report generation script capable of handling a million data points coming from markets around the world in ~30 languages, this script saved ~9 hours of manual work and brought down the manpower required from 4 to just 1
  • Launched e2e ML service, which enables a naive user to train a basic classification model with their custom dataset, without any knowledge of ML
  • Received 4 recognitions/awards for my work in R&D
ResearchProject ManagementMachine Learning

Education

Georgia Institute of Technology

Masters — Computer Science

Aug 2021Dec 2022

Veermata Jijabai Technological Institute (VJTI)

Bachelor of Technology - BTech — Computer Science

Jan 2015Jan 2019

Shri Vile Parle Kelvani Mandals Mithibai College of Science

Junior College — Science

Jan 2013Jan 2015

Dr. S. Radhakrishnan Vidyalaya

School

Jan 2001Jan 2013

Stackforce found 100+ more professionals with Machine Learning & Distributed Training

Explore similar profiles based on matching skills and experience