Ved Prakash profile photo 1 on Tal

Ved Prakash

Principal Software Engineer - AI/ML

Bengaluru

12+ years of work experience

Atlassian
Aug 20224 y 1 moPresent

Bengaluru, Karnataka, India

AI/ML Engineer executing e2e across ML, AI, data and backend - Built, evaluated and rolled out context engine MCP using FastMCP+FastAPI across Atlassian repos to reduce tokens, latency and imptove accuracy by replacing inefficient tool calling trajectories across large monorepos in Atlassian, empowers ~10 engineer team as foundational layer without worrying about pieces like auth, feature gating, analtics and evaluation and agents-repo integration. - Implemented document ingestion-retrieval-evaluation RAG pipeline to write compliant code across Atlassian products and rolled out as one of the context MCP tool - Wrote and maintain latency optimized inference services for text/code embedding, reranking models as part of context infra services, comes from deep platform/infra experience from pre-LLMs era. - Built mcp bench to evaluate & iterate context mcp across agents like codex/claude code/cursor and rovodev(internal) Coding Agent (Codelassian later packaged as RovodevIDE) - Built inline code completion context evaluation and integrated with VSCode & IntelliJ IDE agents - Built low latency, cost optimized hybrid code completion model inference using vLLM & llama.cpp - HuggingFace model deployments to AWS EC2 & Sagemaker for various embedding use cases. - Owned backend to Integrate streaming llm apis and RAG pipeline for semantic code search. Predictive Test Selection at Scale ( 20K weekly builds) - Trained SVD+XGBoost models and deployed with FastAPI to skip tests in CI branch builds to reduce build time and infra cost - Built MLOPs, CI-CD automations using bitbucket pipelines - Trained classical ML models like XGBoost using SKLearn Data residency for OpsGenie/JSM Multi-region data migration service moving 5 million DynamoDB records in 30 minutes. Designed and implemented event-driven architecture using Kotlin/SpringBoot micro-services and AWS SNS, SQS, Kinesis & Lambda ensuring application consistent migration across multi-region DynamoDB and ElasticSearch.
NetApp Public Cloud Services
Jan 20211 yr 7 mosJul 2022

Bengaluru, Karnataka, India

Implemented reserved capacity offering of Azure NetApp Files, and customer managed encryption key for NetApp Volume Service in GCP, while working on multi-tenant storage+network control plane orchestration backed by NetApp's proprietary OS ONTAP on data plane. Tech Stack: Golang, C#, RabbitMQ, Kubernetes, MariaDB, Azure, GCP, AWS, Docker
Microsoft
Oct 20155 yrs 4 mosJan 2021

- Wrote big data pipeline which processed 100s of TBs of data to show entity pane (or its attribute) among millions of knowledge base entities with 2 more members. Implemented the streaming pipeline to propagate upstream fact changes to Bing under 1s. The batch data pipeline include web scale knowledge graph pruning & transformation, data mining, pattern extraction, writing freshness & relevance rankers, entity intent & sub intent classifiers, feature modeling to rank knowledge entities against a query on Bing. - Implemented video analytics and transcripts migration from Teams to Microsoft Stream
Oracle
Jun 20132 yrsMay 2015

Bengaluru, Karnataka, India

Oracle RDBMS - Worked with backup & recovery team (RMAN) of Oracle database, first hand experience of database internals, backups, replication for a RDBMS.

Ved’s AI Stack

anthropic apis
anthropic apis
claude code
claude code
codex
codex
cursor
cursor
hugging face
hugging face
ollama
ollama

Education

National institute of technology Calicut

Bachelor of Technology (B.Tech.), Computer Science

National institute of technology Calicut

20094 yrs2013

Join Ved and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

Ved Prakash on Tal

Headline

Principal Software Engineer - AI/ML

About

ML + Backend Engineer, built coding agents and context engine using enterprise data ie. internal repos, docs, worked across domains like ML, AI, data engeering and backend services

Location

Bengaluru

AI stack

anthropic apis

claude code

codex

cursor

hugging face

ollama

Work experience

Senior Machine Learning Engineer at Atlassian, Aug 2022 - Present · 4 y 1 mo, Bengaluru, Karnataka, India, AI/ML Engineer executing e2e across ML, AI, data and backend - Built, evaluated and rolled out context engine MCP using FastMCP+FastAPI across Atlassian repos to reduce tokens, latency and imptove accuracy by replacing inefficient tool calling trajectories across large monorepos in Atlassian, empowers ~10 engineer team as foundational layer without worrying about pieces like auth, feature gating, analtics and evaluation and agents-repo integration. - Implemented document ingestion-retrieval-evaluation RAG pipeline to write compliant code across Atlassian products and rolled out as one of the context MCP tool - Wrote and maintain latency optimized inference services for text/code embedding, reranking models as part of context infra services, comes from deep platform/infra experience from pre-LLMs era. - Built mcp bench to evaluate & iterate context mcp across agents like codex/claude code/cursor and rovodev(internal) Coding Agent (Codelassian later packaged as RovodevIDE) - Built inline code completion context evaluation and integrated with VSCode & IntelliJ IDE agents - Built low latency, cost optimized hybrid code completion model inference using vLLM & llama.cpp - HuggingFace model deployments to AWS EC2 & Sagemaker for various embedding use cases. - Owned backend to Integrate streaming llm apis and RAG pipeline for semantic code search. Predictive Test Selection at Scale ( 20K weekly builds) - Trained SVD+XGBoost models and deployed with FastAPI to skip tests in CI branch builds to reduce build time and infra cost - Built MLOPs, CI-CD automations using bitbucket pipelines - Trained classical ML models like XGBoost using SKLearn Data residency for OpsGenie/JSM Multi-region data migration service moving 5 million DynamoDB records in 30 minutes. Designed and implemented event-driven architecture using Kotlin/SpringBoot micro-services and AWS SNS, SQS, Kinesis & Lambda ensuring application consistent migration across multi-region DynamoDB and ElasticSearch.

Member of Technical Staff IV at NetApp Public Cloud Services, Jan 2021 - Jul 2022 · 1 yr 7 mos, Bengaluru, Karnataka, India, Implemented reserved capacity offering of Azure NetApp Files, and customer managed encryption key for NetApp Volume Service in GCP, while working on multi-tenant storage+network control plane orchestration backed by NetApp's proprietary OS ONTAP on data plane. Tech Stack: Golang, C#, RabbitMQ, Kubernetes, MariaDB, Azure, GCP, AWS, Docker

Data Scientist II at Microsoft, Oct 2015 - Jan 2021 · 5 yrs 4 mos, - Wrote big data pipeline which processed 100s of TBs of data to show entity pane (or its attribute) among millions of knowledge base entities with 2 more members. Implemented the streaming pipeline to propagate upstream fact changes to Bing under 1s. The batch data pipeline include web scale knowledge graph pruning & transformation, data mining, pattern extraction, writing freshness & relevance rankers, entity intent & sub intent classifiers, feature modeling to rank knowledge entities against a query on Bing. - Implemented video analytics and transcripts migration from Teams to Microsoft Stream

Member Technical Staff at Oracle, Jun 2013 - May 2015 · 2 yrs, Bengaluru, Karnataka, India, Oracle RDBMS - Worked with backup & recovery team (RMAN) of Oracle database, first hand experience of database internals, backups, replication for a RDBMS.

Education

Bachelor of Technology (B.Tech.), Computer Science, National institute of technology Calicut