Deepak Batra profile photo 1 on Tal

Deepak Batra

Senior Software Engineer @ Google | Machine Learning, Big Data Infrastructure, Java, Algorithms

Bengaluru

15+ years of work experience

Google
Jun 20197 yrs 1 moPresent

Bengaluru

* Drove the strategic organizational migration of the continuous training Ads-wide ML platform from TensorFlow to JAX, optimizing infrastructure for 10,000+ daily models and delivering annualized cost efficiencies of $XXM. * Spearheaded the design and implementation of cross-functional metadata interoperability standards and Source-of-Truth (SoT) lineage tracking for all AdsML assets achieving $XM in yearly resource savings. * Served as a founding architect for the GCS Insights team, defining the technical roadmap and core reporting infrastructure to enable enterprise customers to derive actionable analytics from exabyte-scale storage data.
Microsoft
May 20192 mosJun 2019

Microsoft's Big data Team (Cosmos)
Myntra
Mar 20181 yr 2 mosApr 2019

Bengaluru Area, India

▪ Leading the architecture and design for a scalable and efficient Self-Serve Data Ingestion Framework being built using Gobblin, Kafka and Kafka Connect, which can ingest data from variety of sources like Databases, RestAPIs, Message Queues, Filesystems et al. ▪ Part of the architecture and design team of an in-house ML Serving Platform being built using open-source projects like ML-Flow, Kubernetes ▪ Led the architecture, design and development for a scalable, configurable and efficient Self-Serve Data Processing engine (streaming + batch) built over Apache-Spark with Spark ML capabilities ▪ Working on a Data discoverability project which would serve as a Data Catalogue for all data present in the Data Lake (S3) to be connected to the self-serve Ingestion and Processing Frameworks
Myntra
May 20161 yr 11 mosMar 2018

Bangalore

▪ Re-Architected the entire clickstream pipeline for scaling, efficiency and configurability with almost zero data loss guarantees. Read about various big-data tools which can be used to solve the present issues, and shortlisted a few with help of the team after their evaluation. Came up with an architecture which is being put into use piece by piece in the pipeline. ▪ Lead the design of the real-time pipeline serving data to dashboards.
Skyhigh Networks
Dec 20141 yr 4 mosMar 2016

Bangalore

. Led the work of improving the existing Data Analytics pipeline based on Oozie, MapReduce and HBase etc. . Designed and implementation enhancements i.e. performance and new requirements in the existing pipeline.
Guavus
Mar 201410 mosDec 2014

Gurgaon, India

. Worked on the design of an in-house distributed cache over Apache-spark. . Led the design to move the already built up generic config-driven MR execution engine to Apache-Spark.
Guavus
Mar 20122 yrs 1 moMar 2014

Gurgaon, India

. POCs over Apache-Spark to be used in the next Analytics Platform to be built in Guavus. . Worked on an efficient MR job scheduling engine on top of Cascading 2.0 with a configuration as its input. It generates a dependency DAG of jobs and schedules/executes them efficiently on a cluster. Proposed and discussed configuration driven framework designs with technical advisors and developed it from scratch with complete ownership. . Part of a three member Map-Reduce team, involved in the design and development of a pluggable, performance optimized MR framework and the solution side code processing huge data
Guavus
Aug 20118 mosMar 2012

Gurgaon, India

. As part of Report Generation Engine dev. team, worked on the report generation part using eclipse-BIRT. Wrote a multi-threaded, concurrency aware engine, which generated reports by querying a distributed cache and emit reports in various formats. . Worked on the memory footprint benchmarking of an in-house distributed cache

Deepak’s AI Stack

hugging face
hugging face
langchain
langchain
langgraph
langgraph
ollama
ollama

Education

Netaji Subhas Institute of Technology

B. E, Computers

Netaji Subhas Institute of Technology

20054 yrs2009

Senior Secondary, Science Non-medical

Mira model school

20022 yrs2004

Join Deepak and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

Deepak Batra on Tal

Headline

Senior Software Engineer @ Google | Machine Learning, Big Data Infrastructure, Java, Algorithms

Location

Bengaluru

AI stack

hugging face

langchain

langgraph

ollama

Work experience

Senior Software Engineer at Google, Jun 2019 - Present · 7 yrs 1 mo, Bengaluru, * Drove the strategic organizational migration of the continuous training Ads-wide ML platform from TensorFlow to JAX, optimizing infrastructure for 10,000+ daily models and delivering annualized cost efficiencies of $XXM. * Spearheaded the design and implementation of cross-functional metadata interoperability standards and Source-of-Truth (SoT) lineage tracking for all AdsML assets achieving $XM in yearly resource savings. * Served as a founding architect for the GCS Insights team, defining the technical roadmap and core reporting infrastructure to enable enterprise customers to derive actionable analytics from exabyte-scale storage data.

Senior Software Engineer at Microsoft, May 2019 - Jun 2019 · 2 mos, Microsoft's Big data Team (Cosmos)

Associate Architect - Analytics at Myntra, Mar 2018 - Apr 2019 · 1 yr 2 mos, Bengaluru Area, India, ▪ Leading the architecture and design for a scalable and efficient Self-Serve Data Ingestion Framework being built using Gobblin, Kafka and Kafka Connect, which can ingest data from variety of sources like Databases, RestAPIs, Message Queues, Filesystems et al. ▪ Part of the architecture and design team of an in-house ML Serving Platform being built using open-source projects like ML-Flow, Kubernetes ▪ Led the architecture, design and development for a scalable, configurable and efficient Self-Serve Data Processing engine (streaming + batch) built over Apache-Spark with Spark ML capabilities ▪ Working on a Data discoverability project which would serve as a Data Catalogue for all data present in the Data Lake (S3) to be connected to the self-serve Ingestion and Processing Frameworks

Lead Technology at Myntra, May 2016 - Mar 2018 · 1 yr 11 mos, Bangalore, ▪ Re-Architected the entire clickstream pipeline for scaling, efficiency and configurability with almost zero data loss guarantees. Read about various big-data tools which can be used to solve the present issues, and shortlisted a few with help of the team after their evaluation. Came up with an architecture which is being put into use piece by piece in the pipeline. ▪ Lead the design of the real-time pipeline serving data to dashboards.

Member Of Technical Staff at Skyhigh Networks, Dec 2014 - Mar 2016 · 1 yr 4 mos, Bangalore, . Led the work of improving the existing Data Analytics pipeline based on Oozie, MapReduce and HBase etc. . Designed and implementation enhancements i.e. performance and new requirements in the existing pipeline.

Lead Technology at Guavus, Mar 2014 - Dec 2014 · 10 mos, Gurgaon, India, . Worked on the design of an in-house distributed cache over Apache-spark. . Led the design to move the already built up generic config-driven MR execution engine to Apache-Spark.

Sr. Software Engineer at Guavus, Mar 2012 - Mar 2014 · 2 yrs 1 mo, Gurgaon, India, . POCs over Apache-Spark to be used in the next Analytics Platform to be built in Guavus. . Worked on an efficient MR job scheduling engine on top of Cascading 2.0 with a configuration as its input. It generates a dependency DAG of jobs and schedules/executes them efficiently on a cluster. Proposed and discussed configuration driven framework designs with technical advisors and developed it from scratch with complete ownership. . Part of a three member Map-Reduce team, involved in the design and development of a pluggable, performance optimized MR framework and the solution side code processing huge data

Software Engineer at Guavus, Aug 2011 - Mar 2012 · 8 mos, Gurgaon, India, . As part of Report Generation Engine dev. team, worked on the report generation part using eclipse-BIRT. Wrote a multi-threaded, concurrency aware engine, which generated reports by querying a distributed cache and emit reports in various formats. . Worked on the memory footprint benchmarking of an in-house distributed cache

Education

B. E, Computers, Netaji Subhas Institute of Technology

Senior Secondary, Science Non-medical, Mira model school