Harshit Yadav profile photo 1 on Tal

Harshit Yadav

SDE-2 @Navi | Intern @Samsung R&D | IIT KGP Maths & Computer Science

Bengaluru

3+ years of work experience

Navi
Jul 20242 yrs 2 mosPresent

Bengaluru South, Karnataka, India

Specialized in high-performance ML systems, LLMs, and distributed computing. My key contributions include architecting a multi-agent on-call automation system using LangGraph and a RAG pipeline with Qdrant, which leverages semantic search on historical incidents for proactive resolution suggestions. I have also developed a multi-version, scalable model inference service featuring a Cython-optimized package, achieving 5ms latency and reducing AWS costs by 60% through dynamic batching and asynchronous gRPC calls via the Triton Inference server. Additionally, I engineered a high-throughput, distributed data aggregator in Kotlin to ensure seamless data accessibility for ML/DL models.
Navi
Jul 20231 yr 1 moJul 2024

Bengaluru, Karnataka, India

Recognized with the performance-based Navi Spearhead award from a team of 40 data science professionals. Specialized in software development and machine learning infrastructure. Architected a model serving orchestrator in Golang that resulted in an 80% reduction in AWS costs and a 95% drop in latency. My work included implementing robust model metadata management using Postgres and Redis, and building data pipelines with S3 and Kafka. I also engineered a scalable, generic FastAPI service for ML/DL models and designed automated deployment pipelines using GoCD, Kubernetes, and GitHub Actions to improve deployment efficiency and model reliability.
Samsung R&D Institute India
May 20222 mosJun 2022

Bengaluru, Karnataka, India

Developed a Hindi Chatbot for Samsung Finance Service using RASA with the NLU training datasets and domains. Created multiple custom actions to facilitate multilingual support, utilizing a language-agnostic core server, Rasa NLU servers via Entity Mapping.
SymPy, Open-Source Organization
Jan 20214 mosApr 2021

Designed, developed, and documented the DomainMatrix module, optimizing matrix operations for increased efficiency. Implemented new methods and functionalities for SymPy modules like Matrix, Core, Polys, Derivatives, and Integrals.
Leoforce
May 20192 mosJun 2019

Hyderabad, Telangana, India

Worked as a Front-End JavaScript developer on a live project based on the web application of a management system for the employee through which feedback is gathered. Developed various dashboards and MBO templates in Production using JavaScript, HTML, SQL and resolved various Production bugs related to JavaScript and SQL in the application.
GitHub
@hyadav2k

Harshit’s AI Stack

antigravity
antigravity
codex
codex
cursor
cursor

Education

Indian Institute of Technology, Kharagpur

Master's degree, Mathematics and Computer Science

Indian Institute of Technology, Kharagpur

20185 yrs2023

Higher Secondary School

SR Public Sr Sec School, Kota

20162 yrs2018

Join Harshit and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Harshit Yadav on Tal

Headline

SDE-2 @Navi | Intern @Samsung R&D | IIT KGP Maths & Computer Science

About

Software Development Engineer, holding an advanced degree in Mathematics and Computer Science from the Indian Institute of Technology, Kharagpur, with a proven record of architecting, developing, and deploying high-performance, scalable systems. Expertise spans the complete software development lifecycle, from initial conceptualization and strategic design to robust implementation, rigorous testing, and ongoing maintenance, with a particular specialization in engineering sophisticated solutions within machine learning systems, Large Language Models (LLMs), and distributed computing environments. Notable accomplishments underscore a commitment to innovation and operational efficiency. These include the architecture of a complex multi-agent on-call automation system for proactive incident resolution; the development of a high-throughput model inference service that achieved 5ms latency while reducing AWS expenditures by 60% through dynamic batching; and the engineering of a versatile, scalable FastAPI service supporting over 40 distinct machine learning and deep learning models. Advanced proficiency in Go and Python is complemented by extensive, hands-on experience with a comprehensive suite of modern development technologies, including Docker, Kubernetes, LangChain, LangGraph, and Qdrant, enabling the construction of resilient and economically viable software solutions.

Location

Bengaluru

AI stack

antigravity

codex

cursor

Work experience

Software Engineer - 2 at Navi, Jul 2024 - Present · 2 yrs 2 mos, Bengaluru South, Karnataka, India, Specialized in high-performance ML systems, LLMs, and distributed computing. My key contributions include architecting a multi-agent on-call automation system using LangGraph and a RAG pipeline with Qdrant, which leverages semantic search on historical incidents for proactive resolution suggestions. I have also developed a multi-version, scalable model inference service featuring a Cython-optimized package, achieving 5ms latency and reducing AWS costs by 60% through dynamic batching and asynchronous gRPC calls via the Triton Inference server. Additionally, I engineered a high-throughput, distributed data aggregator in Kotlin to ensure seamless data accessibility for ML/DL models.

Software Engineer at Navi, Jul 2023 - Jul 2024 · 1 yr 1 mo, Bengaluru, Karnataka, India, Recognized with the performance-based Navi Spearhead award from a team of 40 data science professionals. Specialized in software development and machine learning infrastructure. Architected a model serving orchestrator in Golang that resulted in an 80% reduction in AWS costs and a 95% drop in latency. My work included implementing robust model metadata management using Postgres and Redis, and building data pipelines with S3 and Kafka. I also engineered a scalable, generic FastAPI service for ML/DL models and designed automated deployment pipelines using GoCD, Kubernetes, and GitHub Actions to improve deployment efficiency and model reliability.

Software Engineer Intern at Samsung R&D Institute India, May 2022 - Jun 2022 · 2 mos, Bengaluru, Karnataka, India, Developed a Hindi Chatbot for Samsung Finance Service using RASA with the NLU training datasets and domains. Created multiple custom actions to facilitate multilingual support, utilizing a language-agnostic core server, Rasa NLU servers via Entity Mapping.

Open Source Developer at SymPy, Open-Source Organization, Jan 2021 - Apr 2021 · 4 mos, Designed, developed, and documented the DomainMatrix module, optimizing matrix operations for increased efficiency. Implemented new methods and functionalities for SymPy modules like Matrix, Core, Polys, Derivatives, and Integrals.

Web Development Intern at Leoforce, May 2019 - Jun 2019 · 2 mos, Hyderabad, Telangana, India, Worked as a Front-End JavaScript developer on a live project based on the web application of a management system for the employee through which feedback is gathered. Developed various dashboards and MBO templates in Production using JavaScript, HTML, SQL and resolved various Production bugs related to JavaScript and SQL in the application.

Education

Master's degree, Mathematics and Computer Science, Indian Institute of Technology, Kharagpur

Higher Secondary School, SR Public Sr Sec School, Kota

Public links

github: https://github.com/hyadav2k