Smriti Nisha

Software Engineer

Hyderabad, Telangana, India9 yrs 10 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Designed low-latency infrastructure for Azure AI.
Led latency reduction for Bing's Copilot Voice Search.
Developed automation tools reducing operational costs.

Stackforce AI infers this person is a SaaS-focused Software Engineer with expertise in AI and automation.

Contact

Skills

Core Skills

C#Azure AiSpeech RecognitionJava

Other Skills

AJAXAudio Codec IntegrationCreative WritingDesign PatternsHTMLJ2EE Web ServicesJUnitJavaScriptLatency OptimizationNon-profit VolunteeringOracle DatabaseOracle SQL DeveloperPublic SpeakingReal-time MonitoringSQL

About

A team-player and quick learner with keen interest in coding, algorithms and solving challenging problems. I am passionate to learn new things.

Experience

9 yrs 10 mos

Total Experience

3 yrs 3 mos

Average Tenure

7 yrs 2 mos

Current Experience

Microsoft

2 roles

Software Engineer 2

Promoted

Aug 2020 – Present · 5 yrs 10 mos

Azure AI Platform Team
Designed and built end-to-end low-latency inference infrastructure for Azure Model Router with support for multi-model deployment on a single endpoint, maintaining <50ms p99 latency and controlled GPU utilization.
Developed a comparison evaluation pipeline to benchmark model router against competitors and also a real-time monitoring dashboard to track model performance, latency, and service health.
Optimized inference code, improving throughput and latency by 4x for long context. Optimized inference notebooks to run 10x faster, making customer adoption easy in private preview of Model Router.
Speech Recognition | Voice Search
Led latency reduction efforts for Copilot Voice Search on Bing, implementing audio codec integration and slow-network-aware routing to achieve up to 1.7s reduction in Speech to Text and Text to Speech latency globally.
Took ownership of E2E speech model training for 4 Indic locales (bn-IN, mr-IN, kn-IN, gu-IN), achieving significant WERR improvements and ~20x reduction in ghost word errors.
Created high-quality data preparation pipeline for Speech to text training, reducing the time for training data readiness by ~5x.
Developed a slow-network classifier, resulting in 258ms (16.6%) improvement in TTS AudioReady metrics at P75 for voice search on Copilot for slow network users.

C#Azure AILatency OptimizationReal-time MonitoringSpeech Recognition

Software Engineer

Mar 2019 – Jul 2020 · 1 yr 4 mos

Translation Answer on Bing
Shipped major Translation Answer features including Dark Mode, Soft Keyboard support for 50+ languages, Image Integration using Getty/Bing APIs for 10K+ entities, Answer Grounding, Auto-Suggest and Phrasebook in translation answer, which gave major gain in customer engagement and +5 point global gain in WPSBS metrics against competitors..

Amazon

Software Development Engineer

Jun 2018 – Mar 2019 · 9 mos · Greater Hyderabad Area

Transportation Technology Team
Built an automation tool to streamline carrier onboarding, significantly reducing manual effort and operational costs.
Developed backend services for Manifest generation and printable document workflow for Amazon’s Transportation Fulfilment systems.

Oracle india pvt. ltd

Application Developer

Jul 2016 – Jun 2018 · 1 yr 11 mos · Hyderabad, Telangana, India

Cloud Product Master (CPM)
Built and maintained backend services for Cloud Product Master (CPM), the core metadata and revenue tracking system for Oracle Cloud products.
Developed role-based UI and integrated access approval workflow and email notifications. Enabled mass Excel uploads via Apache POI to streamline product data operations.

Syntel

Software Developer Intern

May 2015 – Jul 2015 · 2 mos · Greater Chennai Area

Project Details- UNIT TESTING AUTOMATION FRAMEWORK : A framework that eliminates the effort of manual creation and maintenance of JUnit test cases in the test driven environment, saves the valid test case in the form of XML and also revalidates the existing functionalities in the modified scenarios.
Developed Core Engine in JAVA to invoke the method of a class dynamically with the passed parameters using JAVA Reflection API.
Stored the valid test cases in the form of XML using JAXB framework.
Used the same test case to revalidate the functionality in the modified scenario.
Developed UI using JAVA Swing toolkit.

Dakshana foundation

Summer Analytics Intern

May 2014 – Jun 2014 · 1 mo · Pune

Dakshana Foundation is US based Non Profit Organization having its operational office in India at Pune.
Project Details:
1. Reduced the cost of operation and increased participation by 20 percent changing the location of 98 test centers spread across India.
2. Developed algorithm to gauge scholars in Grand Dakshana Selection Test .
3. Increased efficiency of operation by building quality assessment tools to track progress of scholars and teachers.