Abhishek Soni

Software Engineer

Delhi, India4 yrs 11 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

Led development of LLM-powered experiences at Amazon.
Achieved top placements in national hackathons.
Improved wake-word detection systems at Samsung.

Stackforce AI infers this person is a Software Engineer specializing in AI-driven solutions for consumer applications.

Contact

Skills

Core Skills

AwsJavaArtificial Intelligence (ai)Cloud ApplicationsMachine LearningSoftware Development

Other Skills

Business RequirementsCode ReviewSpring BootCommunicationOral CommunicationSoftware Development Life Cycle (SDLC)Academic Support ServicesSoftware InfrastructureMockitoAmazon CloudWatchAmazon EC2AWS LambdaSoftware DesignAndroidSDKTypeScript

About

As a passionate and results-driven Software Development Engineer with a strong background in electronics, communication, and software engineering, I specialize in building scalable, high-performance solutions. Currently at Amazon, I have led multiple impactful projects, including developing conversational experiences powered by Large Language Models for Prime Video, streamlining cloud deployment pipelines, and mentoring interns to success. I thrive on challenges, having previously improved wake-word detection systems, optimized test processes, and developed real-time notification systems at Samsung. Proficient in Java, C++, and various AWS technologies, I bring strong problem-solving abilities and a collaborative mindset. I have a proven track record in hackathons, securing 1st and 3rd place at national competitions with certifications from Amazon and Samsung, I aim to keep growing while delivering value through innovative technology. Skills: * Programming Languages: Java, C++, Python, TypeScript, JavaScript, Matlab * Frameworks: Node.js, React.js * Cloud Platforms: AWS, Azure (ADO, Logic Apps, Sentinel, Runbooks, Function Apps, Cognitive Services & Azure ML), Apache Kafka * Databases: PostgreSQL, SQL * Machine Learning: TensorFlow, PyTorch, Scikit-learn * Unit Testing Frameworks: Jest, GTest, Mockito

Experience

4 yrs 11 mos

Total Experience

2 yrs 5 mos

Average Tenure

3 yrs 9 mos

Current Experience

Amazon

Software Engineer

Sep 2022 – Present · 3 yrs 9 mos · Bengaluru, Karnataka, India

1st Place (National) in AVS Hackathon for the project ”Alexa for Emergencies” and 3rd Place (National) in Alexa LLM Hackathon with project ”Amazon Lens”
Wake-word engine improved for 2 stage detectors, improving latency by 1ms and reducing FRR by 0.5 percent for ABI Devices
Independently mentored an intern from project initiation to completion, resulting in a solution that reduced binary initial test time from 2 hours to 5 minutes.
Led the development of a conversational experience powered by Large Language Models (LLMs) for Prime Video search and playback, successfully rolled out to over 10 million LG WEBOS TV devices.
Successfully on boarded a team of 3 onto xApp and played a key role in delivering its first implementation for the Android segment.
Created an AWS code pipeline for Orange, reducing deployment time from 4 hours of manual work to a streamlined, automated process for partner deployments

Business RequirementsCode ReviewAWSJava

Samsung electronics

Software Engineer

Jul 2021 – Sep 2022 · 1 yr 2 mos · Delhi, India

TEM score and test case coverage of project increased from 13 percent to 70 percent on java spring boot using Mockito.
Created a robust notification system using Java Spring Boot that delivers immediate Samsung Knox messages based
on predefined escalation policies, improved incident response times by 40 percent across relevant teams.
Achieved Samsung SWC professional certification within first 5 months of joining.
Configured SAML authentication on Kibana instance using x-pack and Microsoft ADSSO as identity provider. Updated
the entire ELK dev cluster version.

Business RequirementsCode ReviewJavaMachine Learning

Leadingindia.ai

Machine Learning Intern

May 2020 – Jun 2020 · 1 mo · Delhi, India

My project aimed on developing a software to recognize, predict and identify Arrhythmia from a dataset consisting of 30-minute ECG samples of 48 subjects.
We were able to classify arrhythmia samples of patients with an accuracy of 95.55% by using and optimizing recognized machine learning models.

Choti si khushi

Tutor

May 2018 – Jul 2018 · 2 mos · Delhi, India

We were a team of 4 dedicated to teaching poor children between the age of 8-12 yo at a nearby community park. We covered concepts from basic science and maths to advanced exam preparation for specific students.

CommunicationOral Communication