Jul 2023 – Jul 2024 · 1 yr · Gurugram, Haryana, India · Remote

• Part of Core Data Team: managed accuracy and coverage of 20B+ datapoints.
• Architected, designed and developed a full-stack solution capable of supporting multi-fold concurrent users, ultimately contributing to a company revenue of over $1.5 million / year.
• Built foundational infrastructure enabling 50+ integrations in the Web App, boosting AEs’
• outreach speed by 1.5X (Refer product release below).
• Created migration jobs handling 20B+ data points, resulting in 7.5% enhancement in phone number accuracy.
• Wrote multi-fold scripts to automate enrichment data exercises, that ultimately decreased turn around time from 5 days to couple of minutes.
• Part of hiring team - helped 10+ engineers get hired every 6 months.
• Technologies used: Mongo, Redis, Ruby on Rails, Snowflake, ES, GCP, K8s, Docker, Sentry

May 2022 – Jul 2023 · 1 yr 2 mos · Hyderabad, Telangana, India

• Part of Ad Platforms Team: managed Ad booking portal for Apple News, Stocks and App Store.
• Tech Stack: Kafka, Splunk, Jenkins, Python, Consul, Java, OracleDB, Internal tools

Sep 2020 – Mar 2022 · 1 yr 6 mos

• Designed, built and scaled a full stack service to detect stylistic inconsistencies within a text.
• Impact: Established a new revenue stream for the company (refer public release of the product down below).
• Responsible for bootstrapping multiple projects, handling it from scoping phase to wrapping it up in a production ready environment.
• Tech Stack: AWS, EC2, DynamoDB, Sentry, CloudWatch, S3, Lambda, API Gateway

May 2020 – Aug 2020 · 3 mos · Remote

• Being an early member of the project - Implemented and scaled a Federated Learning package for simulating decentralized training on common biomedical datasets, by using Data Version Control system, Differential Privacy & WebHooks.
• Orchestrated federated ML training on a network of EC2 instances.
• Tech Stack: DVC library, PySyft, Docker, TF-Federated, TF-Encrypted, GitHub APIs

Jan 2020 – Aug 2021 · 1 yr 7 mos

• Using in-silico simulations to find potential drug candidates for treating COVID-19.
• Tech Stack: RDKit, Tensorflow, Docker, Flask, Heroku, Semantic UI
• Learn more at - https://bion-ai.herokuapp.com/

Jan 2020 – Jul 2020 · 6 mos

• Built a Web-App, for 3D visualization of drug-like compounds and proteins.
• Tech Stack: RDKit, 3Dmol.js, Docker, Flask, Heroku
• Learn more at https://visualize-2020.herokuapp.com/

Aug 2019 – Dec 2019 · 4 mos · University of Texas, El Paso, US

• Architected, designed and developed a full stack service (from scratch) for predicting subcellular locations & for measuring Anti-SARS-CoV-2 activity, given a chemical molecule. Implemented a consensus ML approach for the same.
• Implemented multiple RESTful APIs and built 40 ML classifiers (trained on clusters). Evaluated the models across 5 metrics, and achieved 99.8% accuracy having F1-score of 0.94, for this task.
• Implemented a web-crawler and scraped chemical data from 5 databases.
• Conducted ML workshops and sessions for 10+ students, helped them in their tenpole final year projects.
• Tech Stack: Docker, scikit-learn, Graph CNN, Selenium, BeautifulSoup, Dask, Bootstrap, Semantic UI

May 2019 – Jul 2019 · 2 mos · Remote

• Led the development of an ML platform, with Ensembl’s Genome Analysis Team,
• for correctly characterizing, naming and identifying lncRNA genes.
• Generated 30 custom features and benchmarked a selection of ML algos. Accuracy ≈ 90%.
• Technologies used: Biopython, Pyfasta, gffPandas, RESTful APIs, scikit-learn

Dec 2018 – Jan 2019 · 1 mo

• Developed a web-based tool for calculating Drug Toxicity Index, which tells us how much toxicity a drug will have, compared to other standard values.
• Link for the web-tool : https://all-tool-kit.github.io/Web-Tool.html
• Learn more here - https://pubs.rsc.org/en/content/articlelanding/2019/tx/c8tx00261d#!divAbstract

Aug 2018 – Dec 2018 · 4 mos

• Developed an Android app which recognizes patterns and detects plant diseases.
• Implemented watershed algorithm, contours and U-Net segmentation.
• Tech Stack: Tensorflow, Java, Alex-Net, Open-CV
• Learn more at https://github.com/vermasrijan/PlantDiseaseDetectionApp

Jun 2018 – Dec 2018 · 6 mos · Bengaluru Area, India

• Responsible for building LSTM models (from scratch), using Multispectral/Hyperspectral satellite images, for crop yield
• prediction (on 666 Indian districts) and finding illegal mines. Accuracy ≈ 88%.
• Technologies used: Tensorflow, QGIS, NDVI Masking, Google Earth Engine

May 2018 – Jul 2018 · 2 mos · Bangalore, India

• Led the development of a Neural Machine Translator, having an accuracy of ~85%, for a sports analytical firm, Messy Fractals. Used seq2seq architecture in PyTorch framework so as to translate from English to other reg. languages.
• Wrote a web crawler to scrape, clean and assimilate text data as inputs to NLP model.
• Visualized word embedding clusters (on tensor board) of Hindi and English text using t-SNE and K-Means algorithm.
• Built a Language Model using word embeddings to predict the next word / set of words for Hindi Text.
• Used Mozilla TensorFlow Deep Speech to implement a Speech2Text model for converting sports commentary to its text format.
• Tech Stack: Tensorflow, PyTorch, t-SNE & K-Means, NLTK

Jun 2017 – Jul 2017 · 1 mo · Gurugram, Haryana, India

• Collaborated with web developers and led the improvement of Search Engine Rank of the website with various SEO measures.
• Built a chatbot using motion.ai framework which actively interacts with users and helps them create customized nutrition plans and monitors their progress. The Chatbot helps reduce the time spent by the nutritionists with the users by up to 30%.
• Piloted an AI driven tool that helps Ophthalmologists accurately predict the particular stage of an eye disease and differentiate between the different retina-based eye diseases.
• Tech Stack: Python, Machine Learning, Tensorflow, Java, SEO optimization

Dec 2016 – Jan 2017 · 1 mo · Gurugram, Haryana, India

• Responsible for representing the ML team in designing, implementing and architecting MLOps flow across 3 critical backend services.
• Revamped login and payment page which reduced user friction by 15% and increased user traffic by ~18.5%.
• Tech Stack: Swift