Feb 2023 – Mar 2025 · 2 yrs 1 mo · Remote

• TL;DR - Part of the ML Infrastructure team; Building storage, ingestion and compute systems for machine learning
• Leading development of ML Compute Platforms, based on a distributed compute infrastructure.
• Led search & feature discoverability projects for ML Infra metadata; setting up data pipelines, modeling indexes, materialization and ingestion to search index db.
• Actively helped data science teams to migrate long running batch inference jobs to ray, reducing job run time by 77% (4.5x improvement).
• Scoped and carried out caching and query optimizations for feature store (ML storage platform)
• Built a scheduler and data ingestion service for feature store
• Created an SDK library for ML platform services.
• Built feature store integrations with other platform services
• Carried out PoCs for enhancing MLOps capabilities

May 2022 – Jan 2023 · 8 mos · Remote

• Building an in-house Feature Store (ML storage system) from scratch as a part of our ML Platform; implemented storage and metadata layer
• Helped revamp a Data Quality Rule Engine, giving ability to use pre-defined checks, evaluate custom expressions, and orchestrate them through trino and spark jobs via Kubernetes APIs.
• Wrote Lexers, Parsers to build ASTs (Abstract Syntax Trees) and AST Evaluators in Golang for evaluating DQ Validation Expressions

Jan 2023 – Jan 2023 · 0 mo

Sep 2022 – Sep 2022 · 0 mo

Jan 2021 – May 2022 · 1 yr 4 mos · Gurugram, Haryana, India

• TL;DR - Leading Storage Infrastructure, Search, and ML
• Being the first Data Scientist / Software Engineer at Koverhoop, I have been working on excitingly tough problems, in Insurance, and Real Estate domains.
• Leadership:
• Led the development of the entire Storage Infrastructure from Database Replication, Real-time Stream Processing (Kafka based pipelines), to ML Pipelines to democratize data, set up BI & KPIs, and deploy models.
• Led development to build the in-house observability system based on open telemetry, replacing managed solutions, overall saving costs.
• Managed DevOps, from K8s to CI/CD pipelines and MLOps for klient.ca
• Architecture and Design for a novel open-source, multi-level text annotation system, Astronaut, leading a team of five.
• Having a say in company's direction.
• I actively hire for Infra, Data, and Platform teams.
• Projects:
• Engineering stream-processing pipelines to denormalize relational data on-the-fly, and load it into elasticsearch to enable a search capability
• Building models to extract information from Legal Insurance Contracts, to automate lengthy form filling process, saving thousands of man-hours annually.
• Leading development of a novel, multi-level text annotation tool, Astronaut.sh.

Jul 2019 – Dec 2020 · 1 yr 5 mos · Gurugram, Haryana, India

• Projects:
• Designed the backend of a comprehensive reporting micro-service, that automates the creation of well-defined excel reports, and presents interactive dashboards to business users.
• Built a group insurance price estimation model to reduce the conversion time from prospect to opportunity by 60%.
• Created algorithms, using OpenCV to remove duplicate real-estate listings to maintain data quality and enhance the user experience

Jan 2020 – Jan 2020 · 0 mo

Jan 2019 – May 2019 · 4 mos · Gurugram, Haryana, India

• Analysing sentiment scores from tweets related to relevant stocks and predicting the sentiment impact on the market for the next day. Used state-of-art RNNs to evaluate scores.

May 2017 – Aug 2017 · 3 mos · New Delhi Area, India

• Helped build traveldglobe’s content platform. The services I developed helped in increasing the pagerank of traveldglobe.com pages on search engines.
• Created services for five main functionalities: Most searched words, Missing keywords in content, Content Keyword Extraction, Google Rank Analysis, Plagiarism Detection.