Arpit Kumar

Software Engineer

Delhi, India4 yrs 1 mo experience

Most Likely To SwitchHighly Stable

Key Highlights

Reduced indexing costs by $3M through optimization.
Managed API handling over 10M queries daily.
Improved email search click rates from 2% to 8%.

Stackforce AI infers this person is a Software Engineer with expertise in SaaS and Machine Learning.

Contact

arpitkumar@microsoft.com LinkedIn

Skills

Core Skills

C#SearchDecision TreeReact.jsNode.jsKubernetesPython

Other Skills

.NET FrameworkApache KafkaBootstrapCSSComputer VisionData StructuresDatabasesDeep LearningDjangoDockerEquityEquity ResearchEquity ValuationExpress.jsFull-Stack Development

Experience

Microsoft

2 roles

Software Engineer 2

Aug 2024 – Present · 1 yr 7 mos · Hyderabad, Telangana, India · On-site

• Spearheaded end-to-end integration of messages as suggestions in Teams search bar, adding 1M extra impressions on the search bar. Worked seamlessly with multiple development teams ensuring quality standards within aggressive timelines.

C#.NET FrameworkDecision TreeSearch

Software Engineer 1

Jul 2022 – Aug 2024 · 2 yrs 1 mo · Hyderabad, Telangana, India · On-site

Drastically reduced the cost of indexing ($3M) by optimizing indexing for users based on their search activity, parallelly ensuring minimal loss in search relevance. Was awarded with Special Performance Award for the same, given to 4 people per year in the Org.
Worked on A/B testing page ranker for email search results in Outlook using Gradient-boosted Decision Tree, improving the click rate for email from 2% to 8%.
Single-handedly managed SearchHistory API, serving user’s search history as search suggestions to over 10 clients (Teams, Outlook, Edge etc) with the API handling over 10M queries per day and maintaining availability of 99.9%.
Improved latency of SearchHistory API from 450ms to 200ms by adding an additional layer of index caching in the microservice and improved the click-rate of search history suggestions from 9% to 15% by reducing indexing of user’s query from 2mins to 10ms.

C#.NET FrameworkDecision TreeSearch

Urban company

Software Engineer Intern

Jan 2022 – Jun 2022 · 5 mos · Gurugram, Haryana, India · Hybrid

Working with team handling bookings fulfillment revamping how each customer request is processed.
Redesigned the bookings page of the app and updated the backend (database, micro-service) to reduce latency of the page from 900ms to
250ms. Worked with React Native for frontend and Node.js, Express and MongoDB for the backend.

React.jsExpress.jsTypeScriptJavaScriptNode.js

Cisco

Technical Undergraduate Intern

May 2021 – Jul 2021 · 2 mos · Bengaluru, Karnataka, India

Worked with the platform team for enterprise networking, designing OS for DNAC. Built Test Automation Services and APIs for various services offered by Maglev. Worked with Kubernetes, Jenkins, Python, Java and built APIs and test applications for Kafka, PostgresSQL, ElasticSearch along with
testing the services for sizing and stress.

Apache KafkaPostgresDockerKubernetesPython

Codeninja.ai

Machine Learning Engineer

Sep 2020 – Nov 2020 · 2 mos

Built an accounting system for the company to extract information from scanned invoices using Computer Vision and ML(Graph Neural Networks and tesseract).
Used object detection(YOLOv3 and FasterRCNN) along with image-2-text libraries(pytesseract), numpy and openCV.

Tamil nadu health system project (tnhsp)

Data Analyst

May 2020 – Jul 2020 · 2 mos

Successfully built an alert system using Machine Learning
models and GIS mapping for cluster detection of
Tuberculosis and Covid-19 from scratch in 3 months.

Cortex technologies

Computer Vision Engineer

Jan 2020 – Mar 2020 · 2 mos · Pilani, Rajasthan

Worked on creating a mind-controlled prosthetic arm for amputees using EEG data mapped through OpenBCI and a camera-controlled object detection.
Used openCV and Intel RealSense camera for object detection and classification

Csir-ceeri

Student Researcher

Sep 2019 – Dec 2019 · 3 mos · Pilani, Rajasthan

Bulit a model for recognition of various different types of plant species based on their leaves. Used state of the art CNN structures using transfer learning. Used keras, tensorflow and numpy for the same