Sep 2021 – Dec 2021 · 3 mos · Dubai, United Arab Emirates

• Working with UnboxAI - we do AI, Deep Learning, and mathematics with some of the biggest organizations and data sets in the world.
• Developing state-of-the-art natural-language-understanding inspired AI models for consumption at scale - using Transformers and unsupervised learning.
• Domain: NLU, NLP, unsupervised learning, contrastive learning, vision, Transformers, GPT3, ImageGPT, ELECTRA, BERT, CLIP.

Sep 2020 – Jul 2021 · 10 mos

• One of the founding engineers at Jina AI - an open source neural search framework for semantic search using Deep Learning. Jina is listed in CBInsights top 100 AI startups 2021 within a year of its inception.
• Engineering – worked across different components of the search workflow – encoding, indexing and ranking of documents at scale using state of the art neural search models.
• Engineering leadership – working with the CTO, Head of Product and Developer Relations across a breadth of responsibilities, open source maintainer and reviewer of Jina’s hub and examples repositories.
• Speaker at Stanford ML Sys Seminar Episode 21 (https://youtu.be/WniuYVHqL80), PyCon USA Poster Presentation 2021, NeurIPS 2020 (India chapter), PyCon US 2021 Poster Presentation
• Co-author of the Industry Day presentation at ECIR (European Conference on Information Retrieval) on ‘Building Decentralised Neural Search Systems’

May 2019 – Sep 2020 · 1 yr 4 mos

• Working on software development (full-stack) on productivity tools for Trusted Interactions and Platforms for different verticals like:
• Safe-browsing
• Google Shopping
• Google Ads.
• This is a part of the Counter Abuse Technology Core engineering team of Google, impacting developer productivity by several folds.
• Core software developer contributor to Google Assistant's LightHouse Zurich project as part of Google's 20% initiative.

Nov 2018 – May 2019 · 6 mos · India

• Worked with the Epoch team at Nutanix, contributing to the cloud monitoring application's development
• Worked primarily on Alerts microservice in Golang used for anomaly detection and alerting outages for timeseries metric data (based on Druid)

Feb 2018 – Sep 2019 · 1 yr 7 mos

• Contributing to the 'Negative Grants' project to introduce 'deny' privilege command in MariaDB.
• https://mariadb.org/about/staff/rutuja-surve/

Aug 2017 – Nov 2017 · 3 mos

• Contributed to the OPNFV Bottlenecks Project under the Linux Foundation for their Euphrates release by:
• Designing and developing a monitoring module for stress tests of VNFs in Python using Prometheus for monitoring
• Used CollectD for collecting metrics, Grafana for plotting results and Docker for containerizing. Automated the entire application
• Invited to present at Open Networking Summit, North America 2018, held at Los Angeles

Jul 2017 – Oct 2018 · 1 yr 3 mos · Bengaluru Area, India

• Worked on features in the Data Ingestion and Cloud teams of Infoworks, a Big Data start-up that automates data warehousing on Hadoop
• Contributed to exporting enterprise data to BigQuery on Google Cloud Platform, the cloud solution having major impact on customer business for the start-up
• Explored and delivered features using various Google Cloud services like Cloud Storage, Compute Engine, Dataproc

Jun 2017 – Jul 2017 · 1 mo · Bengaluru, Karnataka, India

• Participated as a Code-Sprint Intern with the Open Web Application Security Project (OWASP), a regular GSoC organization
• Built a Machine Learning web-server log analysis tool using Python libraries Numpy, SciPy, Sci-Kit Learn, K-Modes
• It accepts Web Server (Apache) logs as input and provides output as a determination of requests that are considered 'attacks'
• Used different supervised and unsupervised algorithms like Clustering, Decision Tree, Naive Bayes, etc. for classification

Jan 2017 – Jun 2017 · 5 mos · Pune

• Worked on Nvidia’s Video Encoder using Parallel Computing and Image Processing techniques to boost up the speed of encoding. Contributed by: • Improving the encoding by compressing or encoding certain low frequency regions of an image as lossy and high frequency regions as loss-less. This speeds up the encoding and preserves the quality of the image streamed. • The project was completed in two phases: one being the detection of the visible regions of windows and the second being recognition of high frequency content in the regions detected in a series of live screenshots, performed on the fly. • This remote desktop solution, (CUDA C++ and OpenCV) can identify high frequency content regions in a continuously streaming video or desktop.

Aug 2016 – Dec 2016 · 4 mos · Bangalore

• Worked on the Book-Series and Video Games page of Amazon and contributed by:
• Ensuring that multiple author's information was being displayed for books with multiple authors
• Migrating the source of author images to a globalized datapath endpoint, making it independent of market-place
• Translating the Book-Series page to Spanish, holding a direct impact on a million plus customers
• Working on a Machine Learning algorithm for fetching 'What Should I Read Next?' recommendations in Ruby on Rails
• Modifying the media detail-page template for the 'Title block' feature of the Video Games page, to make it more appealing to customers

May 2016 – Aug 2016 · 3 mos · Pune

• Participated as one of the 41 internationally selected interns in the May-August 2016 round of Outreachy hosted by Software Freedom Conservancy, with
• special support from Red Hat and GNOME foundation. Made the following contributions to Mozilla:
• Designed and developed a content-process management tool that can track real time memory usage across each process in the Firefox browser
• The project helped scale up and tune the number of content processes that Firefox starts or uses, keeping memory measurement in mind
• The tool will be accessible through the advanced settings of the Firefox browser and it reports the RSS and USS memory performance in an eventtriggered
• manner for all open tabs. It is platform-independent and works well on operating systems like Windows, Linux and Mac OS X